The Syntax of Korean Anaphora: an Experimental Investigation

Home , Anaphora (linguistics), Bound variable pronoun, Cha (Korean surname), Sloppy identity

The syntax of Korean anaphora: An experimental investigation

by Kyeong-min Kim

M.A., Kyung Hee University, 2011 B.A., Kyung Hee University, 2008

Thesis Submitted in Partial Fulﬁllment of the Requirements for the Degree of Doctor of Philosophy

in the Department of Linguistics Faculty of Arts and Social Sciences

c Kyeong-min Kim 2019 SIMON FRASER UNIVERSITY Spring 2019

All rights reserved. However, in accordance with the Copyright Act of Canada, this work may be reproduced without authorization under the conditions for “Fair Dealing.” Therefore, limited reproduction of this work for the purposes of private study, research, education, satire, parody, criticism, review and news reporting is likely to be in accordance with the law, particularly if cited appropriately. Approval

Name: Kyeong-min Kim Degree: Doctor of Philosophy (Linguistics) Title: The syntax of Korean anaphora: An experimental investigation Examining Committee: Chair: Yue Wang Professor

Chung-hye Han Senior Supervisor Professor

Nancy Hedberg Supervisor Professor

Keir Moulton Supervisor Assistant Professor Department of Linguistics University of Toronto

Christian Guilbault Internal Examiner Associate Professor Department of French

Shinichiro Fukuda External Examiner Associate Professor Department of Linguistics University of Hawai‘i at M¯anoa

Date Defended: December 20, 2018

ii Ethics Statement

iii Abstract

This dissertation investigates the syntactic and interpretative properties of three Korean anaphora, third-person pronouns, VP anaphors (VPAs), and null objects (NOs), using experimental methodologies. There is no general consensus among previous studies regarding whether Korean third-person pronoun ku ‘he’ can be construed as a bound variable. Three interconnected experiments were conducted to explore this issue, and the findings demonstrated that some Korean speakers consistently accepted the quantificational binding of ku, while others consistently did not. This result is highly suggestive of inter-speaker variation in the bound variable construal for ku. Taking into consideration the historical background of ku and its present status, I conclude that child learners of Korean may not receive sufficient evidence regarding ku from the primary language input data. Given this, adopting Han et al.’s (2007) two-grammar hypothesis and Déchaine and Wiltschko’s (2002) pronominal typology, I propose that some speakers randomly acquire FP ku, which complies with the “pronominal grammar”, while others randomly acquire DP ku, which complies with the “demonstrative grammar”. On the basis of the finding that there is inter-speaker variation in the bound variable construal for ku, the present study investigates the syntax of Korean VPAs and NOs. The existing proposals on their syntactic identities can be grouped into ellipsis and pro-form approaches. In two independent experiments designed to diagnose the presence of “hidden” structure within VPAs and NOs, I examined the (un)availability of sloppy readings for VPAs and NOs with antecedents containing ku. Given the standard view that the sloppy reading in ellipsis is due to a pronoun in the ellipsis site being bound, if VPAs or NOs have elided structure that hosts ku, the distribution of sloppy readings for them should correlate with that of quantificational binding of ku. Such a correlation, however, is not expected if they are pro-forms that do not host elided material (and thus not ku). The correlation was found in the experiment for NOs, but not in the experiment for VPAs. Based on these findings, I claim that VPAs are uniform, un-analyzable pro-forms, while NOs are derived from ellipsis, anaphora that have a fully-fledged structure.

Keywords: third-person pronouns; bound variable construal; quantiﬁcational binding; inter-speaker variation; VP anaphors; null objects; ellipsis; pro-form; sloppy reading; Korean

iv Acknowledgements

This dissertation would not have been possible without the invaluable input, guidance, and support received from a number of great individuals. I am very pleased and honored to have this opportunity to express my heartfelt thanks to them. First and foremost, my deepest appreciation is extended to my senior supervisor, Dr. Chung-hye Han. Throughout my years of doctoral study at SFU Linguistics, she has been the best supervisor one could have ever imagined: always accessible, always quick and ready to respond even to “half-baked” ideas or arguments and always encouraging. It goes without saying that this dissertation would not have become a reality without her exceptional supervision, scholarly guidance, timely advice and endless patience. I also wish to acknowledge the financial support received from SSHRC 435-2014-0161 to her. Kyoswunim! Cwueysonghako cinsimulo kamsatulipnita. Conkyenghapnita! Words are not adequate to express sincere gratitude towards my two supervisors, Dr. Keir Moulton and Dr. Nancy Hedberg, for their perspicacious advice and constructive criticisms that have helped to enrich and strengthen many aspects of this dissertation. I am also enor- mously indebted to Dr. Christian Guilbault and Dr. Shinichiro Fukuda, my internal examiner and external examiner, for carefully reading my dissertation and giving me insightful comments and suggestions for revision. I would also like to give my sincere gratitude to Dr. Yue Wang for being the chair of my defense, and am particularly grateful for her course, Research Techniques, which helped me build solid foundations in experimental methodologies. My thanks must also be expressed to my MA supervisor, Dr. Jong-Bok Kim, who initially led me into the world of syntax and encouraged me to pursue a doctorate degree. I also owed many thanks to the past and present administrative staff at the Linguistics Department, Rita Parmar, Silvana Di Tosto, Judi Levang, Christie Carlson, Debra Purdy Kong and Ray Haleem, who always kindly provided me with great service and assistance on non-academic problems. I would also like to gratefully acknowledge the financial support from SFU Linguistics through Graduate Fellowships and various Teaching Assistantships. I have had the privilege of being able to present earlier versions of this dissertation at various conferences and workshops, and have greatly benefited from the critical feedback to the presentations. Especially valuable and helpful were the Impact of Pronominal Form on Interpretation Workshop, the 19th International Circle of Korean Linguistics Conference,

v the 29th Annual CUNY Conference on Human Sentence Processing, the Workshop on Ex- perimental Approaches to East Asian Languages and the 24th Japanese/Korean Linguistics Conference, where I have profited from the incisive questions, comments and criticisms of Martina Wiltschko, Patrick Grosz, Pritty Patel-Grosz, Maria Carminati, James Hye-Suk Yoon, Michael Barrie, Kyumin Kim, Nayoung Kwon, Jeffrey T. Runner, Jon Sprouse, Kazue Kanno, William O’Grady and Heejeong Ko. Heartfelt thanks also go to my fellow graduate students, Stanley Anonby, Irina Pres- nyakova, Queenie Chan, Pocholo Umbal, Keith Leung, Debopam Das, Lisa Shorten, Amber Blenkiron, Meghan Jeffrey, Lindsay Walker, Kayleigh MacMillan, Emma Mileva and Sylvia Cho, who helped me get through the long, dark, rainy winters in Vancouver. Without their friendship, emotional support and encouragement, the long journey to the completion of this dissertation would have been very lonely and painful. Queenie Chan deserves a special mention, since her painstaking proofreading, critical comments and editorial inputs have improved immeasurably the content and form of this dissertation. I am also grateful to the former and current members of the Experimental Syntax Lab, Trevor Block, Holly Gendron, Sara Williamson, Sander Nederveen, Kazuki Yabe, Tanie Cheng and Sophie Thompson, for listening attentively to many of my practice presentations. Among my friends in Korea, I particularly wish to thank Dr. Seung-Han Lee and Dr. Jungsoo Kim for always being there and ready to share my joys and sorrows. I would like to sincerely thank my parents, Jeonghoi Kim and Hieajeong Kim, and my brother, Seongmin Kim, for their spiritual and emotional support throughout the course of this research and, of course, in my entire life. My appreciation goes as well to my parents- in-laws, Seungjoon Choi and Jihyun Hong, for their unceasing encouragement and trust. Finally, I thank my lovely wife, Heejon Choi, for her endless love, understanding, support and sacrifice over the long haul. Cengmal salanghako komawue! Special thanks go to my dear son, Shawn Sieon Kim, just for being there.

vi Table of Contents

Approval ii

Ethics Statement iii

Abstract iv

Acknowledgements v

Table of Contents vii

List of Tables x

List of Figures xi

List of Abbreviations xiii

1 Introduction 1 1.1 Motivation and objectives ...... 1 1.1.1 Third-person pronouns in Korean ...... 2 1.1.2 VP anaphora in Korean ...... 6 1.1.3 Null objects in Korean ...... 14 1.2 Outline ...... 15

2 Inter-speaker variation in Korean pronouns 17 2.1 Introduction ...... 17 2.2 Theoretical Background ...... 19 2.2.1 Competition-based Approach ...... 20 2.2.2 Constraint-based Approach ...... 23 2.2.3 Structural Approach ...... 25 2.3 Experiment 1 ...... 30 2.3.1 Experiment 1A ...... 31 2.3.2 Experiment 1B ...... 39 2.4 Experiment 2 ...... 44 2.4.1 Methodology ...... 44

vii 2.4.2 Findings and Discussion ...... 46 2.5 General discussions ...... 49

3 The syntax of Korean VP anaphora 60 3.1 Introduction ...... 60 3.2 Experiment 3: Sloppy identity reading in Korean VP anaphora ...... 64 3.2.1 Theoretical background ...... 64 3.2.2 Research question and predictions ...... 69 3.2.3 Methodology ...... 72 3.2.4 Findings ...... 76 3.2.5 Discussion of Experiment 3 ...... 80 3.3 Experiment 4: Overt extraction out of Korean VP anaphora ...... 81 3.3.1 Theoretical background ...... 81 3.3.2 Research question and predictions ...... 87 3.3.3 Methodology ...... 89 3.3.4 Findings ...... 94 3.3.5 Discussion of Experiment 4 ...... 98 3.4 General discussions ...... 99

4 The syntax of null objects in Korean 104 4.1 Introduction ...... 104 4.2 Theoretical background ...... 109 4.2.1 Korean as a radical pro-drop language ...... 109 4.2.2 Ellipsis versus null pronominal ...... 114 4.3 Experiment 5 ...... 122 4.3.1 Research question and predictions of the experiment ...... 122 4.3.2 Methodology ...... 126 4.3.3 Discussion of Experiment 5 ...... 133 4.4 General discussions ...... 136

5 Conclusion 140 5.1 Summary of the major ﬁndings ...... 140 5.2 Contributions ...... 142 5.3 Suggestion for future work ...... 142

Bibliography 145

Appendix A Test sentences from Experiment 1A 162 A.1 Condition 1: Quantiﬁcational-Overt ...... 162 A.2 Condition 2: Quantiﬁcational-Null ...... 164 A.3 Condition 3: Referential-Overt ...... 166

viii A.4 Condition 4: Referential-Null ...... 167

Appendix B Test sentences from Experiment 1B 169 B.1 Condition 1: Simple ...... 169 B.2 Condition 2: Complex ...... 170

Appendix C Test sentences from Experiment 2 171 C.1 Condition 1: Bound-August ...... 171 C.2 Condition 2: Bound-September ...... 171 C.3 Condition 3: Free-August ...... 173 C.4 Condition 4: Free-September ...... 174

Appendix D Test sentences from Experiment 3 177 D.1 Condition 1: VPA-Bound (sloppy identity reading) ...... 177 D.2 Condition 2: VPA-Free (strict identity reading) ...... 182 D.3 Condition 3: Quantiﬁcational-Bound ...... 188 D.4 Condition 4: Quantiﬁcational-Free ...... 192

Appendix E Test sentences from Experiment 4 196 E.1 Condition 1: Simple-Extraction (short distance scrambling) ...... 196 E.2 Condition 2: Simple-NoExtraction ...... 204 E.3 Condition 3: Complex-Extraction (long distance scrambling) ...... 212 E.4 Condition 4: Complex-NoExtraction ...... 221

Appendix F Test sentences from Experiment 5 231 F.1 Condition 1: NullObject-Bound (sloppy identity reading) ...... 231 F.2 Condition 2: NullObject-Free (strict identity reading) ...... 237 F.3 Condition 3: Quantiﬁcational-Bound ...... 242 F.4 Condition 4: Quantiﬁcational-Free ...... 242

ix List of Tables

Table 1.1 Hankamer and Sag’s (1976) two-way classiﬁcation of anaphora . . . . 8

Table 2.1 Test conditions in Experiment 1A ...... 33 Table 2.2 Test conditions in Experiment 2 ...... 45

Table 4.1 Syntactic status of null objects ...... 108 Table 4.2 Syntactic status of null objects ...... 136

x List of Figures

Figure 2.1 Screenshot of a test trial in Experiment 1A ...... 32 Figure 2.2 Mean rates of acceptance and standard errors in Experiment 1A . . 37 Figure 2.3 Distribution of responses in Quantiﬁcational-Overt condition in Ex- periment 1A ...... 38 Figure 2.4 Mean rates of acceptance and standard errors in Experiment 1B . . 42 Figure 2.5 Distribution of responses in Simple and Complex conditions in Ex- periment 1B ...... 43 Figure 2.6 Correlation between mean acceptance rates in Simple and Complex conditions in Experiment 1B ...... 43 Figure 2.7 Mean rates of acceptance and standard errors in Experiment 2 . . 47 Figure 2.8 Distribution of responses in Bound conditions for August and Septem- ber sessions in Experiment 2 ...... 47 Figure 2.9 Correlation between mean acceptance rates in Bound conditions for August and September sessions in Experiment 2 ...... 48

Figure 3.1 Screenshot of a test trial in Experiment 3 ...... 73 Figure 3.2 Mean rates of acceptance and standard errors in Experiment 3 . . . 77 Figure 3.3 Distribution of responses in Quantificational-Bound and VPA-Bound (sloppy reading) conditions in Experiment 3 ...... 78 Figure 3.4 Correlation between mean acceptance rates in Quantificational-Bound and VPA-Bound (sloppy reading) conditions in Experiment 3 . . . 79 Figure 3.5 Screenshot of a test trial in Experiment 4 ...... 90 Figure 3.6 Mean acceptability ratings and standard errors for test conditions in Experiment 4 ...... 95 Figure 3.7 Mean acceptability ratings and standard errors for “non-VPA control” conditions in Experiment 4 ...... 96 Figure 3.8 Mean VPA difference and NonVPA difference scores and standard errors in Experiment 4 ...... 97

Figure 4.1 Screenshot of a test trial in Experiment 5 ...... 126 Figure 4.2 Mean rates of acceptance and standard errors in Experiment 5 . . . 131

xi Figure 4.3 Distribution of responses in Quantiﬁcational-Bound and NullObject- Bound (sloppy identity reading) conditions in Experiment 5 . . . . 131 Figure 4.4 Correlation between mean acceptance rates in Quantiﬁcational-Bound and NullObject-Bound (sloppy identity reading) conditions in Exper- iment 5 ...... 133 Figure 4.5 Mean rates of acceptance and standard errors in Experiment 3 and Experiment 5 ...... 135

xii List of Abbreviations

2 second person neg negative 3 third person nom nominative acc accusative obj object adn adnominal part particle aux auxiliary past past tense comp complementizer pl plural conj conjunction poss possessive dat dative prenom prenominal decl declarative pres present tense def deﬁnite prog progressive dem demonstrative q question det determiner sg singular emph emphasis sub subject gen genitive top topic hon honoriﬁc tr transitive link linking masc masculine

xiii Chapter 1

Introduction

1.1 Motivation and objectives

This dissertation explores the phenomenon of anaphora in Korean. In particular, eﬀorts are made to unravel the syntactic and interpretative properties of three Korean anaphoric devices, (i) third-person pronouns, (ii) verb phrase (VP) anaphora, and (iii) null objects, which are exempliﬁed by the sentences in (1)-(3) below ([e] in (3) indicates a phonologically missing object).1

(1) Third-person pronouns

Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl

‘Everyone drank his beverage.’ (2) VP anaphora

Minswu-ka ku-uy umlyoswu-lul masi-ess-ko, Kiswu-to Minswu-nom he-gen beverage-acc drink-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl

‘Minswu drank his beverage, and Kiswu did so, too.’ (3) Null objects

Minswu-ka ku-uy umlyoswu-lul masi-ess-ko, Kiswu-to [e] Minswu-nom he-gen beverage-acc drink-past-conj Kiswu-also masi-ess-ta. drink-past-decl

‘(lit.) Minswu drank his beverage, and Kiswu drank, too.’

1The Yale Romanization is used in transcribing the Korean data throughout this dissertation.

1 This chapter aims to provide the motivation and the main objectives of the present study on the three anaphoric devices listed above. Before doing so, let us ﬁrst consider the extract from Wasow (1986: 107-108) to look at how the anaphora phenomenon is deﬁned in his work.

“An appropriate place to begin my survey is with an attempt to characterize what anaphora is. I will not be so bold as to offer a definition: a particularly successful empirical investigation may end with a definition, but research in progress typically involves imprecise ‘seat of the pants’ characterizations of the phenomena in question. Loosely, then the study of anaphora deals with pronouns, other pro-forms (e.g., do so), and ellipsis. These constructions share a number of properties, the most obvious being that they derive their interpretations in a context from their association with other elements in the context.”

Although anaphora has been extensively studied in various fields of linguistics for the past decades, its definitions have not been entirely consistent among researchers and, as pointed out by Garnham and Oakhill (1996: 320), any attempt to formulate a complete and fully satisfactory definition of anaphora seems to have not been successful, because of the wide scope and variety of the phenomenon. However, adopting the idea suggested by Wasow (1986: 107-108) above, the present study defines anaphora as the linguistic phenomenon (or process) wherein the interpretation of one (overt or covert) element is in some way dependent on the interpretation of the other element previously given either in the linguistic or extralinguistic context (see also Huang 1994 for a similar definition of anaphora).2

1.1.1 Third-person pronouns in Korean

As is well known, there is a broad range of anaphoric expressions that are legitimately employed by natural language. Among these devices, pronominal anaphora is the one that is probably the most commonly used and has drawn the greatest attention in generative grammar. It is a generally accepted view that pronouns are involved in, at least, two distinct types of anaphoric relations: coreference anaphora and bound variable anaphora. First, a pronoun can enter into coreference when it happens to denote the same referent as some referential nominal expression in the same sentence (or adjacent sentences). In (4), for example, the intended interpretation of the English third-person pronoun him results from being coreferential with the preceding referential DP, future President Andrew Jackson.

(4) On this day in 1806, future President Andrew Jackson kills a man who accused him of cheating on a horse race bet.

2Following the convention in the literature, the term anaphora will also be used to refer to anaphoric expressions in general.

2 (https://www.history.com/this-day-in-history)

On the other hand, bound variable anaphora can be established when a pronoun is construed as a variable that is bound by an operator such as a quantifier. This is illustrated in (5), where the third-person pronoun he is understood as a variable whose value is not fixed, but co-varies with the value that would be assigned to the preceding quantificational DP, every man.

(5) Every man must decide whether he will walk in the light of creative altruism or in the darkness of destructive selﬁshness. (Martin Luther King 1963)

Notably, not all languages include third-person pronouns in their pronominal systems (e.g., for Yidiny, see Dixon 1977; for Basque, see Laka 1996). Moreover, some languages do possess third-person pronouns, but ones that are diﬀerent from English third-person pronouns in terms of the range of possible anaphoric relations. For example, in languages such as Polish and Catalan, third-person pronouns can enter into coreference, but not into bound variable anaphora, as illustrated in the following sets of sentences.3

(6) Polish

a.* Nikt 1 nie wierzy ze on1 jest inteligentny. nobody neg believe.pres comp he be.pres intelligent

‘Nobody1 believes that he1 is intelligent.’

b. Aleksander1 wierzy ze on1 jest inteligentny. Aleksander believe.pres comp he be.pres intelligent

‘Aleksander1 believes that he1 is intelligent.’ (adapted from Barski 2013: 59, ex.(18))

(7) Catalan

a.* Ningú 1 creu que éll1 és antipátic. nobody believe.pres comp he be.pres nasty

‘Nobody1 believes that he1 is nasty.’

b. En Joan1 creu que éll1 és simpatic. det John believe.pres comp he be.pres nice

‘John1 believes that he1 is nice.’ (Picallo 1994: 270, fn. 16, (ia) and (iia))

In (6a) and (7a), the third-person pronouns ón and éll cannot be construed as variables bound by the quantiﬁcational DPs, nikt and ningú ‘nobody’, while in (6b) and (7b), they can be coreferential with the referential DPs, Aleksander and en Joan. Obviously, this

3Catalan is a language spoken in Spain (the Spanish autonomous communities of Catalonia, the Balearic Islands and Valencia), Andorra, and some parts of France and Italy.

3 is in contrast to English third-person pronouns, which can readily receive bound variable interpretations as well as coreferential interpretations. Interestingly enough, Korean third-person pronoun ku ‘he’ displays seemingly a rather unusual anaphoric status, which is distinctive from the Polish and Catalan counterparts as well as the English counterpart.4 To illustrate this point, consider the following two sentences, both of which were extracted from some online postings written by native Korean speakers.

(8) a. Motun salam1-un ku1-ka nwukwu-inci, hananim-kkeyse ku1-ka every person-top he-nom who-Q God-nom.hon he-nom mwues-ul hayngha-tolok kitayha-si-nunci alko-iss-ta. what-acc do-comp expect-hon-Q know-prog-decl 5 ‘Everyone1 knows who he1 is and what God expects him1 to do.’ (https://www.lds.org/general-conference/2000)

b. Motun salam1-un ku1-uy sasaynghwual, kaceng saynghwual, cwuke mit every person-top he-gen privacy family life home and thongsin-ul concwung pat-ul kwuenli-ka iss-ta. correspondence-acc respect receive-prenom right-nom exist-decl

‘Everyone1 has the right to respect for his1 private and family life, (his1)

home, and (his1) correspondence.’ (http://hrlibrary.umn.edu/instree/K-z17euroco.html)

What is noticeable in the above examples is that the pronoun ku is intended to be anaphorically linked to the quantiﬁcational DP, motun salam(-un) ‘everyone’, thus receiving a bound variable reading. However, such a grammatical intuition does not seem to be shared by all native speakers of Korean. For example, of the nine Korean speakers I informally consulted with, six judged that the sentences in (8a) and (8b) are not felicitous under the intended bound variable reading of ku, while the remaining three judged that the same sentences are felicitous under such a reading. Moreover, as will be sketched in Section 2.1 of Chapter 2, there is a clear split in the Korean literature between those who argue that ku can be used to encode a bound variable interpretation (e.g., Im 1987; B. M. Kang 1988; M. Y. Kang 1988; Suh 1990; Koak 2008) and those who argue that it cannot (e.g., Hong 1985; Choe 1988; S. H. Kang 1990; Shim 1993; Kwon et al. 2009).6 If so, this observation might be an indication that there is inter-speaker variation in the bound variable use of Korean

4Korean third-person pronouns also show unique characteristics in terms of the frequency of occurrence in spoken and written contexts. That is, they are frequently used in the written language, but rarely employed in colloquial speech (e.g., O’Grady 1984). See Chapter 2 for the discussion on the historical reason for this context asymmetry and its potential inﬂuences and consequences on children’s acquisition of third-person pronouns in Korean. 5The sentence was originally translated from Fiddler on the Roof (1963), an English play written by Joseph Stein and later turned into a ﬁlm. 6It is commonly assumed in the literature that in contrast to its bound variable use, the third-person pronoun ku can be freely used as a coreferential pronoun. Indeed, all of my informants agreed that the

4 third-person pronouns. However, such a non-trivial possibility has never been reported or addressed in previous studies on Korean anaphora, and thus never been empirically verified or theoretically justified (cf. Marsden 2012; Kweon 2017). One of the main objectives of this dissertation is to contribute to filling this important void in the extant literature.7 In particular, in Chapter 2, on the basis of careful and extensive experimental investigations (Experiments 1A, 1B, and 2), I argue that there is indeed variation across Korean speakers in the availability of bound variable construal for Korean third-person pronouns, and sentences in (i) and (ii) below are felicitous under the intended (intrasentential) coreferential interpretation of ku.

(i) a. Changswu1-nun ku1-uy emeni-lul salangha-n-ta. Changswu-top he-gen mother-acc love-pres-decl

‘Changswu1 loves his1 mother.’ (C. W. Lee 2005: 200, ex.(1a))

b. Minswu1-uy chinkwu-ka ku1-lul ttayli-ess-ta. Minswu-gen friend-nom he-acc hit-past-decl

‘Minswu1’s friend hit him1.’ (Cho & Hong 1988: 32, ex.(4))

(ii) Cwucinwu-ka ku namca1-lul sileha-nun mankhum-ina appa-to ku1-lul Cwucinwu-nom the man-acc dislike-prenom extent-as dad-also he-acc sileha-n-ta. dislike-pres-decl

(lit.) ‘As much as Cwucinwu (a famous Korean journalist) dislikes the man1, Dad dislikes him1, too.’ (http://blog.aladin.co.kr/735181196/popup)

7As a matter of fact, in previous studies on the phenomenon of bound variable anaphora in Korean, much less attention has been devoted to the third-person pronoun ku than the long-distance anaphor caki ‘self’. This is probably due to the fact that in Korean, the use of caki as a (local or non-local) bound variable (e.g., Cho 1996; Storoshenko 2008; Han & Storoshenko 2012) is incomparably more preferable and, thus, more common and widespread than the bound variable use of ku (e.g., O. H. Kim & Kitagawa 2010; Kwon et al. 2009). All of the nine Korean speakers mentioned earlier unequivocally judged that the sentences in (i) and (ii) below, where ku in (8a) and (8b) have been replaced with caki, can readily induce a bound variable interpretation. Most of all, according to those three informants who agreed that ku can be construed as a bound variable, the caki sentences in (i) and (ii) sound far more natural than the corresponding ku sentences in (8a) and (8b).

(i) Motun salam1-un caki1-ka nwukwu-inci, hananim-kkeyse caki1-ka mwues-ul hayngha-tolok every person-top self-nom who-Q God-nom.hon he-nom what-acc do-comp kitayha-si-nunci alko-iss-ta. expect-hon-Q know-prog-decl

‘(lit.) Everyone1 knows who self1 is and what God expects self1 to do.’

(ii) Motun salam1-un caki1-uy sasaynghwual, kaceng saynghwual, cwuke mit thongsin-ul every person-top self-gen privacy family life home and correspondence-acc concwung pat-ul kwuenli-ka iss-ta. respect receive-prenom right-nom exist-decl

‘(lit.) Everyone1 has the right to respect for self1’s private and family life, (self1’s) home, and (self1’s) correspondence.’

In light of this reality, the current research can provide a signiﬁcant contribution to illuminating a more precise and complete picture of Korean (third-person) pronouns.

5 provide an account for why and how the phenomenon of inter-speaker variation arises and exists in Korean.

1.1.2 VP anaphora in Korean

Let us now direct our attention to VP anaphora. In particular, what the present study is concerned with is the so-called do so VP anaphora, a type of predicate anaphora (e.g., Cornish 1986/2005), which can be found, at least, in English and East Asian languages such as Korean, Japanese, and (Mandarin) Chinese.8,9 Consider the examples of English do so and its Japanese and Chinese equivalents, soo-su and zheme-zuo, given in (9) below.10

(9) a. English The data showed that the tropical areas of Southeast Asia, South America

and Africa [VP added the same amount of CO2 into the atmosphere], but they did so for diﬀerent reasons. (https://www.livescience.com/60670-nasa-satellite-reveals-source.html) b. Japanese

A: Taroo-wa [VP sinroo-ni hanataba-o watasi]-ta. Taroo-top bridegroom-dat bouquet-acc hand-past ‘Taro handed a bouquet to the bridegroom.’ B: Hanako-mo soo-si-ta. Hanako-also so-do-past ‘Hanako did so, too.’

8Some researchers, e.g., Hankamer and Sag (1976), Partee and Bach (1984), Huang (2000), use the term VP anaphora (or VP anaphor) to refer to verbal anaphoric items in general, which include null anaphoric VPs, i.e., VP ellipsis. Throughout this dissertation, however, the term VP anaphora is only used to refer to overt anaphoric VPs (e.g., English do so anaphora and do it anaphora), following the convention used in Matsuo and Duﬃeld (2001), Cecchetto and Percus (2006), Y. H. Lee (2012), and Roberts et al. (2013). 9Danish gør det and French le faire have often been assumed to be do so VP anaphors in the literature, based on examples like (i)-(ii) below.

(i) Danish Peter spiser spaghetti om aftenen, men Henrik gør det om eftermiddagen. Peter eats spaghetti in evening, while Henry does it in afternoon ‘Peter eats spaghetti in the evening, while Henry does so in the afternoon.’ (Schøsler 1994: 123, ex. (14d)) (ii) French Jacqueline ne devrait pas nager dans cette mer agitee, mais Georges peut le faire. Jacqueline ne should not swim in this sea rough but Georges can it do ‘Jacqueline shouldn’t swim in that rough sea, but Georges can do so.’ (Cornish 1986/2005: 108, ex. (40b))

However, as can be known from the word-by-word glosses given above, it seems more reasonable to take the Danish and French VP anaphors as equivalents of English do it anaphora. 10In the rest of the dissertation, the term Chinese is used to mean Mandarin Chinese, unless otherwise speciﬁed.

6 (Koizumi 1994: 36, ex.(29)) c. Chinese

Zhangsan zai tushuguan [VP nian shengjing]. Mali zai jiali ye zheme Zhangsan at library read bible Mary at home also so zuo. do ‘Zhangsan read the bible at the library. Mary did so at home, too.’ (Wei & Li 2016: 190, ex.(6a))

In (9a), English do so in the second clause is readily understood as taking on the meaning of the bracketed VP in the preceding clause, and thus the second clause is interpreted as ‘they added about the same amount of CO2 into the atmosphere for diﬀerent reasons’. The Japanese and Chinese VP anaphors in (9b) and (9c) are interpreted in the same way as their English equivalent in (9a), and therefore the second clauses in (9b) and (9c) mean ‘Hanako handed a bouquet to the bridegroom, too’ and ‘Mary read the book at home, too’, respectively. Since the classic and landmark work done by Hankamer and Sag (1976) (and also Sag and Hankamer 1984), the syntactic status of do so VP anaphora has been a topic of some debate in the generative literature, albeit to a much lesser extent than VP ellipsis, which is also a class of predicate anaphora. In order to understand this point in more detail, we need to consider the crux of Hankamer and Sag’s (1976) view on the phenomenon of anaphora. According to their claim, a formal distinction should be drawn between two categories of anaphoric devices, which has become a standard assumption in the subsequent literature on anaphora (cf. Williams 1977; Grimshaw 1979; Tanenhaus & Carlson 1990; Depiante 2000; Johnson 2001; Merchant 2001; Cecchetto & Percus 2006; Houser 2010; cf. Baltin 2012; Kasai 2014; Wei & Li 2016, among many others). The ﬁrst category of anaphoric devices are elliptical constructions, which are derived from fully articulated syntactic forms by deletion under identity with their antecedent forms at “surface” level (i.e., at the PF component), and which look to the linguistic representations (or, more precisely, the logical forms) associated with the antecedents for their interpretation; these anaphoric devices are referred to as surface anaphora (also as deletion anaphora). On the other hand, the second category of anaphoric devices are pro-forms, which are atomic units generated in the deep (or underlying) structure, and which make reference to the discourse or semantic model for

7 Ellipsis (surface anaphors) Pro-forms (deep anaphors) VP ellipsis, Sluicing Do it anaphora Stripping, Gapping, Null complement anaphora, Do so anaphora Sentential it anaphora

Table 1.1: Hankamer and Sag’s (1976) two-way classiﬁcation of anaphora their interpretation; these anaphoric devices are referred to as deep anaphora.11,12 Table 1.1 provides a list of various English anaphoric devices classiﬁed into each of these two categories. As can be inferred from the above discussion, the major factor that distinguishes the two classes of anaphoric devices is the presence/absence of internal syntactic structure. Consider the following examples of English VP ellipsis and do it anaphora, which are representative cases of elliptical anaphora and pro-form anaphora, respectively.13

11It is nowadays a widespread practice to assume Hankamer and Sag’s deep/surface dichotomy for the study of anaphora, but it was a controversial hypothesis when initially proposed. As mentioned in Hankamer and Sag (1976: 394), at the time of the publication of their work, there existed two major, diametrically opposed approaches to anaphora: (i) strict transformational position and (ii) strict interpretive position. According to the ﬁrst approach, “all anaphoric processes are transformations that involve deletion (or con- version to a pro-form) of an underlyingly present, fully lexical segment under conditions of identity with an antecedent segment” (e.g., Ross 1967; Postal 1972). The second view, on the other hand, assumes that “all anaphors (overt or null) are present in underlying representations” and that “the anaphoric relation between an anaphor and its antecedent is assumed to be established by an interpretive rule” (e.g., Wasow 1972; Fiengo 1974). 12In Sag and Hankamer (1984), surface anaphora and deep anaphora are renamed ellipsis and model- interpretive anaphora, respectively. 13Examples of other English anaphoric devices are given below.

(i) Sluicing: Hankamer: Someone’s just been shot. Sag: Yeah, I wonder who. (Hankamer & Sag 1976: 408, ex. (42)) (ii) Stripping: Hankamer: Listen, Ivan, he’s playing the William Tell Overture on the recorder. Sag: Yeah, but not very well. (Hankamer & Sag 1976: 409, ex. (46)) (iii) Gapping: Hankamer: Ivan is now going to peel an apple. Sag: And Jorge, an orange. (Hankamer & Sag 1976: 410, ex. (49)) (iv) Do so anaphora: If you have not yet changed your socks, please do so immediately. (Hankamer & Sag 1976: 415, ex. (70)) (v) Null complement anaphora (NCA): We needed somebody to carry the oats down to the bin, but nobody volunteered. (Hankamer & Sag 1976: 411, ex. (51c)) (vi) Sentential it anaphora: That Betty is claimed to be pregnant doesn’t make it true. (Hankamer & Sag 1976: 421, ex. (101))

As Houser et al. (2007: 183) point out, Hankamer and Sag’s two-way distinction of anaphora is independent of whether an anaphoric item is phonologically null or not. For example, NCA and do it anaphora are a

8 (10) a. VP Ellipsis

Hankamer: I’m going to [VP stuﬀ this ball through this hoop ].

Sag: It’s not clear that you’ll be able to [VP ]. (Hankamer & Sag 1976: 382, ex. (5)) b. Do it anaphora

Hankamer: I’m going to [VP stuﬀ this ball through this hoop ].

Sag: It’s not clear that you’ll be able to [VP do it ]. (adapted from Hankamer & Sag 1976: 382, ex. (4))

In (10a), the VP ellipsis site in the second conjunct (denoted by [VP ]) is assumed to have a full-fledged structure that corresponds to its antecedent in the first conjunct, [VP stuff this ball through this hoop ], throughout the entire syntactic derivation (i.e., in both overt and covert syntax (LF)), although its surface representation is rendered opaque by a phonological deletion process (see Chapter 3 for further discussion). In (10b), on the other hand, the pro-form do it in the second conjunct is assumed to be base-generated as it is, and thus to have no internal structure at any stage of syntactic derivation. Accordingly, it posits no more structure than what must be present for the surface lexical form, i.e., the verb do and the pronoun it. In light of this contrast, the underlying structures of the second conjuncts of (10a) and (10b) can be roughly represented as in (11a) and (11b), respectively.

(11) a. ...

T VP

to V0

V0 PP

V DP through this hoop

stuﬀ this ball null deep anaphor and an overt deep anaphor, respectively, while VP ellipsis and do so anaphora are a null surface anaphor and an overt surface anaphor, respectively (for Spanish and Italian NCAs, see Depiante 2000; for Japanese NCA, see Kasai 2014; for Catalan do it, see Busquets 2018).

9 b. ...

T VP

to V0

V DP

do it

As well summarized in LaCara (2010) and Baltin (2012), an elliptical anaphor has syntactic material inside of it, which can thus interact with (or can be manipulated by) other syntactic processes; by contrast, a pro-form anaphor does not have any internal syntactic structure, which is thus not available to interact with any syntactic processes. In light of this contrast in syntactic accessibility, a number of syntactic phenomena, which require internal structure for inaudible material, have been put forward as diagnostic tests to distinguish elliptical anaphora from pro-form anaphora. The missing antecedent phenomenon (e.g., Grinder & Postal 1971; Hankamer & Sag 1976) and extraction phenomenon (e.g., Depiante 2000; Johnson 2001; Aelbrecht 2010) are two such examples, as illustrated in (12) and (13), respectively.

(12) a. I’ve never ridden a camel, but Ivan has [VP ], and he says it stank horribly. (it = the camel that Ivan has ridden) (Hankamer & Sag 1976: 403, ex. (23)) b.* I’ve never ridden a camel, but Ivan has done it, and he says it stank horribly. (Houser 2010: 10, ex. (24))

(13) a. Although I don’t know who John will visit, I do know who Fred will [VP ]. b.* Although I don’t know who John will visit, I do know who Fred will do it. (Baltin 2012: 382, ex. (2) and (3))

The above grammaticality contrast between the examples of VP ellipsis and do it anaphora can be well accounted for in terms of the presence/absence of internal structure. In (12a), the VP ellipsis site involves “hidden” syntactic structure that can accommodate an element that can serve as the antecedent for the pronoun it, while in (12b), the pro-form do it has no such structure that can house a possible antecedent for the pronoun.14 In (13a), the VP ellipsis site has internal structure, thus being available to provide a base position for a trace

14In (12a), the indeﬁnite DP in the ﬁrst conjunct, a camel, cannot serve as the antecedent for the pronoun it, since it is in the scope of negation and thus cannot introduce a (new) referent into the discourse (Heim 1982), as illustrated in (i).

(i)* I’ve never ridden a camel, and it stank horribly. (Hankamer & Sag 1976: 404, ex. (25))

10 of wh-movement, while in (13b), the do it pro-form does not have any internal syntactic structure in which a trace of wh-movement can sit. Now let us return to the discussion of do so VP anaphora. Since the introduction of the “Deep-Surface” hypothesis of Hankamer and Sag (1976), there have been debates on whether do so VP anaphora in English should be classiﬁed as elliptical anaphora or pro-form anaphora. First, Hankamer and Sag (1976) (see Table 1.1) and other subsequent researchers (e.g., Ward et al. 1991; Fu et al. 2001; Stroik 2001) argue that English do so anaphora is a case of ellipsis, and thus has unpronounced syntactic structure. According to their view, the site of do so involves a more complex and full-ﬂedged structure, whose surface representation is “replaced” (and thus “deleted”) by the overt lexical string do so. This ellipsis analysis has often been supported by the observation that English do so anaphora can license the missing antecedent phenomenon, as illustrated in the sentences in (14).

(14) a. I didn’t ride a camel, but Ivan must have done so, and now our oﬃce is infested with its ﬂeas. (it = the camel that Ivan has ridden) (Hankamer & Sag 1976: 418, ex. (83)) b. Jerry wouldn’t read a book by Babel, but Meryl has done so, and it was pretty good. (it = the book by Babel that Meryl has read) (Johnson 2001: 27, ex. (96))

However, many other researchers argue against the view that English do so is an elliptical anaphor, challenging the diagnostic abilities of the processes that have been proposed to support the presence of its elided structure (e.g., Hardts 1993; Kehler & Ward 1999, 2004; Depiante 2000; cf. Culicover & Jackendoﬀ 2005; Sobin 2008; Houser 2010; cf. Hallman 2013; Bruening 2018).15 They instead claim that English do so anaphora is a VP pro-form, which is a syntactically primitive element that never has internal syntactic structure.16 The observation that wh-movement out of the site of do so is not available, as shown in (15a) and (15b), has been widely taken as evidence supporting the pro-form analysis of English do so anaphora.

On the other hand, the indefinite DP contained within the elided VP structure is not under the scope of negation, as in Ivan has [VP ridden a camel ], and thus can introduce a discourse referent. 15Williams (1977) claims that English do it, which is definitely a pro-form, can invoke the missing antecedent effect on the basis of sentences like (i). (i) John wouldn’t order a new sink, so I did it, and of course it was broken when it arrived. (it = the new sink that I ordered) (Williams 1977: 693, ex. (2)) He also adds that the missing antecedent for the pronoun it can be licensed through pragmatic or semantic inferences (and thus without recourse to any syntactic material). If this is indeed the case, then the missing antecedent phenomenon “cross-cuts” the ellipsis vs. pro-form distinction and, thus, should not be taken as a diagnostic for distinguishing the two classes of anaphora (see also Johnson 2001, Toosarvandani 2009, and Houser 2010 for discussions on the controversy). In light of this consideration, it might be difficult to maintain that the sentences in (14) can support the ellipsis analysis of English do so anaphora. 16See Hallman (2013: 77) for the LF-copying analysis of English do so anaphora, where its interpretation is assumed to be “reconstructed from that of its antecedent by LF copying of its antecedent into the position occupied by do so”.

11 (15) a.* This is the book of which Bill approves, and this is the one of which he can’t do so. (cf. This is the book of which Bill approves, and this is the one of which he

can’t [VP ].) b.* I know which book Max read, and which book Oscar hasn’t done so.

(cf. I know which book Max read, and which book Oscar hasn’t [VP ].) (Johnson 2001: 27, ex. (97a) and (97b))

As can be seen from the discussion thus far, English do so VP anaphora has been prominent in the generative literature, along with other members in the predicate anaphora family (e.g., VP ellipsis and do it anaphora). In contrast, relatively little attention has been devoted to do so VP anaphora in Korean, even though it is broadly used in the language.17,18 The following sentences are naturally occurring examples of the Korean do so anaphora constructions, where kuliha and kulay in (16b) and (16c) are phonological variants of kuleha ‘do so’ in (16a).19

17Do so anaphora in other East Asian languages (e.g., Chinese and Japanese) has also received relatively little treatment in the literature. See, however, Wei and Li (2016) for a recent extensive study on the syntactic structure and derivation of Chinese do so anaphora. 18Very much like its English counterpart, the Korean do so VP anaphora construction has been frequently used as a diagnostics to establish the internal structure within the verb phrase (e.g., to distinguish complements from adjuncts), as illustrated in the sentences in (i)-(iii).

(i) Jim-i chenchenhi pap-ul mek-ess-ko, Susana-nun ppalli kuleha-yess-ta. Jim-nom slowly rice-acc eat-past-conj Susana-top quickly so.do-past-decl ‘Jim ate the rice slowly, and Susana did so (= ate the rice) quickly.’ (ii) Jim-i chenchenhi pap-ul mek-ess-ko, Susana-to kuleha-yess-ta. Jim-nom slowly rice-acc eat-past-conj Susana-also so.do-past-decl ‘Jim ate the rice slowly, and Susana did so (=ate the rice slowly), too.’ (iii)* Jim-i chenchenhi pap-ul mek-ess-ko, Susana-nun chenchenhi kimchi-lul Jim-nom slowly rice-acc eat-past-conj Susana-top slowly kimchi-acc kuleha-yess-ta. so.do-past-decl ‘Jim ate the rice slowly, and Susana did so (= ate) the kimchi slowly.’ (adapted from Shim & Den Dikken 2007: 8, ex.(11))

Note that the VP anaphor kuleha can correspond to either the minimal V0 (the verb mek ‘eat’ and its object complement pap-ul ‘rice-acc’), as in (i), or the larger V0 (the minimal V0 and its adjunct chenchenhi ‘slowly’), as in (ii), while it cannot correspond to the verb alone, as in (iii). The Korean do so anaphora construction has also been used to conﬁrm the bound variable status of the long-distance anaphor caki ‘self’ in the sloppy identity context (e.g., Kim & Yoon 2009; Han & Stroshenko 2012). See Section 3.4 of Chapter 3 for further discussion. 19Do so anaphora is the only VP anaphora that can be found in Korean. Therefore, in the rest of this dissertation, the term Korean VP anaphora is used to mean do so anaphora in Korean.

12 (16) a. Senkyosa-tul-un Ceycwungwon-i senkyopyengwon-i missionary-pl-nom Ceycwungwon-nom missionary.hospital-nom twue-ess-tako sayngkakha-yess-ko, ilpanin-tul-to become-past-comp think-past-conj ordinary.person.pl-also kuleha-yess-ta. so.do-past-decl ‘The missionaries thought that Ceycwungwon became a missionary hospital, and ordinary people did so (= thought that Ceycwungwon became a missionary hospital), too.’

(http://storage.iseverance.com/muobj/add-file/basicclass/3704) b. Uysang-i mom-ul ilukhye anca-ss-ko, Wonhyo-to Uysang-nom body-acc move.up sit-past-conj Wonhyo-also kuliha-yess-ta. so.do-past-decl ‘Uysang sat up, and Wonhyo did so (= sat up), too.’ (http://www.leeinseong.pe.kr/211) c. Chik-un tanhohi kokay-lul ce-ess-ko, na-to kulay-ss-ta. Chik-top flatly head-acc shake-past-conj I-also so.do-past-decl ‘Chik flatly shook his head, and I did so (= flatly shook my head), too.’ (https://books.google.ca/books?id=UtOpAwAAQBAJpg)

There have been two competing approaches to the syntactic identity of Korean VP anaphora: (i) pro-form approaches (e.g., Bae & Kim 2012; M. K. Park 2013), in which VP anaphors are analyzed as base-generated verbal pro-forms, and (ii) ellipsis approaches (e.g., Cho 1996; Son 2006; Ha 2010; Madigan 2015; M. K. Park 2015), in which VP anaphors are analyzed as the results of phonological deletion (or replacement) of a full-ﬂedged VP. One of the main objectives of this dissertation is to contribute to this debate. So far as I am aware, no attempt has yet been made to experimentally investigate the syntax of Korean VP anaphora. In light of this, I conducted two experimental studies designed to diagnose the presence/absence of internal syntactic structure within Korean VP anaphora (Experiment 3 and Experiment 4). More speciﬁcally, I examined whether the Korean VP anaphors can license sloppy identity readings and whether they can permit overt extraction. In Chapter 3, based on the results obtained from these two experiments, I argue that they are instances of pro-form anaphora. Regarding the availability of a sloppy identity reading for an anaphoric item, a number of researchers have taken it as a problematic diagnostics (e.g., Depiante 2000; Houser 2010; Merchant 2013a, 2013b), due to the fact that just like the missing antecedent phenomenon, it cross-cuts the ellipsis vs. pro-form distinction (see Section 3.2 of Chapter 3 for further discussion). However, I demonstrate in Chapter 3 that, with the “great help” of the third-person pronoun ku ‘he’, the availability of a sloppy reading can be used as a reliable diagnostic to identify the syntactic status of Korean VP anaphora.

13 1.1.3 Null objects in Korean

In East Asian languages such as Korean, Chinese, and Japanese, object arguments (as well as subject arguments) are freely allowed to be left unexpressed.20 This is illustrated in (17)- (19) below, where the phonologically null objects are indicated by [e] and their meanings are readily recovered.

(17) Korean A: Chelswu-ka sakwa-lul mek-ess-e. Chelswu-nom apple-acc eat-past-decl ‘Chelswu ate an apple.’ B: Yenghuy-to [e] mek-ess-e. Yenghuy-also eat-past-decl ‘Yenghuy ate an apple, too.’ (J. S. Kim 2012: 3, ex.(7))

(18) Chinese A: Zhangsan bu xihuan shucai. Zhangsan not like vegetables ‘Zhangsan doesn’t like vegetables.’ B: Mali ye bu xihuan [e]. Mary also not like ‘Mary doesn’t like vegetables, either.’ (adapted from S.W. Kim 1999: 25, ex.(1))

(19) Japanese A: Bill-wa tegami-o suteta. Bill-top letter-acc discarded ‘Bill discarded a letter.’ B: John-mo [e] suteta. John-also discarded ‘John discarded a letter, too.’ (adapted from Oku 1998: 3, ex.(7))

The syntactic status of these null objects has long been one of the extensively debated issues in the literature of East Asian languages. The existing proposals can be largely grouped into two camps: (i) null pro-form approaches (e.g., Park 1997; Hoji 1998; Bae & Kim 2012), in which null objects are analyzed as base-generated empty pronominals (pros), and (ii) ellipsis approaches (e.g., Otani & Whitman 1991; Oku 1998; S. W. Kim 1999; Saito 2007;

20Null objects are also found in Romance languages, although they seem somewhat diﬀerent from the counterparts of East Asian languages in terms of the range of possible interpretations. See, for instance, Rizzi (1986) and Farrell (1990) for null objects in Italian and Brazilian Portuguese, respectively.

14 Funakoshi 2016), in which null objects are analyzed as the results of phonological deletion of a full-ﬂedged (DP or VP) structure.21 This dissertation aims to contribute to this long- standing debate. An experimental study (Experiment 5) was conducted to examine the availability of sloppy identity readings for Korean null objects in order to diagnose the presence/absence of internal syntactic structure within them, in a similar way as in the investigation of the syntax of Korean VP anaphora in Experiment 3.22 In Chapter 4, based on the results obtained from this experiment, I argue that Korean null objects are instances of ellipsis.

1.2 Outline

In this chapter, the motivation and the main objectives of the present study have been addressed. This section provides an overview of the chapters that follow. Chapter 2 explores the bound variable status of Korean third-person pronoun ku ‘he’. I present three experimental studies designed to inspect the possibility of inter-speaker variation in the availability of bound variable construal for ku. Experiment 1A examines whether ku can be construed as a bound variable. Experiment 1B examines whether there is a difference between the bound variable construal of ku in a single clause and across clauses. Experiment 2 examines whether Korean speakers exhibit the same judgment over time on the bound variable construal of ku. On the basis of the findings obtained from Experiment 1A, Experiment 1B, and Experiment 2, I aim to determine whether there exists inter-speaker variation regarding the availability of bound variable construal for ku and, if so, to provide a principled explanation for why and how the variation phenomenon arises and exists in Korean. Chapter 3 explores the syntactic identity of VP anaphora in Korean. I present two experimental studies designed to identify the presence/absence of “hidden” syntactic structure within Korean VP anaphora. Building upon the findings regarding the interpretative status of the pronoun ku, Experiment 3 examines whether the distribution of sloppy reading for VP anaphora follows from the distribution of bound variable reading of ku. Experiment 4 examines whether overt extraction out of VP anaphora is available in both the short distance and long distance scrambling contexts. On the basis of the findings obtained from Experiment 3 and Experiment 4, I aim to determine whether Korean VP anaphora is an instance of VP ellipsis or VP pro-form. Chapter 4 explores the syntactic identity of null objects in Korean. I present an experimental study designed to identify the presence/absence of “hidden” syntactic structure

21The ellipsis approaches, in turn, can be divided into two camps: (i) argument ellipsis analysis, in which null objects are viewed as the results of an ellipsis operation called argument ellipsis (or NP/DP ellipsis), and (ii) V-stranding VP ellipsis analysis, in which null objects are viewed as the results of VP ellipsis after verb raising. See Section 4.1 of Chapter 4 for further discussion. 22I also aim to demonstrate that the pronoun ku ‘he’ plays a crucial role in using the availability of a sloppy reading as a reliable diagnostics to identify the syntactic nature of Korean null objects.

15 within null objects in Korean. In a similar way as in Experiment 3, Experiment 5 examines whether the distribution of sloppy reading for null objects follows from the distribution of bound variable reading of ku. On the basis of the ﬁndings obtained from Experiment 5, I aim to determine whether null objects in Korean are instances of ellipsis or null pronouns. Chapter 5 provides a summary of the ﬁndings of this study and concludes with some remarks and future work.

16 Chapter 2

Inter-speaker variation in Korean pronouns

2.1 Introduction

Pronouns are generally known to be interpreted as a bound variable when they are anaphorically linked to a Wh-phrase or a quantiﬁed expression, as in (1) and (2).1

(1) a. Who dislikes his boss? b. Which man did you say dislikes his boss? (2) a. No man should mistreat his friends. b. Every man thinks he is lucky. (Lasnik & Stowell 1991: 687, ex.(1))

The above English sentences can readily receive a bound variable interpretation where the pronoun his or he is understood as a variable bound by the Wh- or quantiﬁed matrix subject, as well as a deictic (or referential) interpretation where the pronoun refers to some male individual in the discourse (e.g., Paul). The distinction between the two interpretations in (2b) may be roughly represented as follows.

(3) a. bound variable reading: ∀x[man(x) −→ think(x, lucky(x))] [(every x : x is a man) x thinks x is lucky] b. deictic reading: ∀x[man(x) −→ think(x, lucky(y))] [(every x : x is a man) x thinks y is lucky]

1Portions of this chapter were published in Patrick Grosz and Pritty Patel-Grosz, eds., The Impact of Pronominal Form on Interpretation (Berlin: De Gruyter Mouton, 2016), 349-373.

17 Interestingly enough, and in contrast to their English counterparts, there is no general consensus among previous studies as to whether Korean third-person pronouns ku ‘he’ and kunye ‘she’ can be construed as a bound variable.2 That is, for some researchers, including the author, the Wh-questions and the quantiﬁcational sentences in (4) and (5) are acceptable under a bound variable reading of ku (e.g., Im 1987; B. M. Kang 1988; M. Y. Kang 1988; Hoji 1990; Suh 1990; Noguchi 1997; Chung 1999; Koak 2008), while for others, the same sentences are not acceptable under such a reading (e.g., Hong 1985; Choe 1988; S. H. Kang 1990; J. R. Kim 1992; Shim 1993; O. H. Kim & Kitagawa 2010; Kanno 1997; Lim 1998; N. K. Kang 2001; Kwon et al. 2009; Choi 2013; Song 2013; Y. H. Kim 2016).3

(4) a. Enu haksayng-i ku-uy kapang-ul ilepeli-ess-ni? which student-nom he-gen bag-acc lose-past-Q ‘Which student lost his bag?’ (Y. H. Kim 2016: 48, ex.(4)) b. Nwu-ka ku-ka Mary-lul bo-ass-ta-ko malha-yess-ni? who-nom he-nom Mary-acc see-past-decl-comp say-past-Q ‘Who said that he saw Mary?’ (Choi 2013: 540, ex.(51)) (5) a. Nwukwunka-ka ku-uy chinkwu-lul paypanha-yess-ta. someone-nom he-gen friend-acc betray-past-decl ‘Someone betrayed his friend.’ (Chung 1999: 134, ex.(17a)) b. Motwu-ka chayksang wuy-ey ku-uy chayk-ul kacyeka-ss-ta. everyone-nom desk on-loc he-gen book-acc take-past-decl ‘Everyone took his book on the desk.’ (S. H. Kang 1990: 127, ex.(32)) c. Amwuto Chelswu-ka ku-lul coaha-n-ta-ko an anyone Chelswu-nom he-acc like-pres-decl-com neg mit-nun-ta. believe-pres-decl ‘No one believes that Chelswu likes him.’ (Shim 1993: 230, ex.(13a)) d. Kak salam-i ku-ka chencay-i-ta-ko sayngkakha-n-ta. each person-nom he-nom genius-be-decl-comp think-pres-decl ‘Each person thinks that he is a genius.’ (B. M. Kang 1988: 421, ex.(14))

2It is to be assumed that all the arguments and conclusions made for ku ‘he’ can be equally well applied to kunye ‘she’, unless otherwise speciﬁed. Thus, kunye will not be mentioned hereinafter. 3In contrast to the bound variable reading, it is commonly agreed that the pronoun ku can readily induce a deictic reading in sentences such as (4) and (5).

18 As pointed out by Oosterhof (2008: 84), “conflicting judgments about the well-formedness of the same sentence type” in the literature may be indication that the relevant linguistic phenomenon is “subject to inter-speaker variation”. If there is indeed variation among native speakers of Korean regarding the availability of a bound variable construal for ku, at least two issues immediately surface. First, the arguments and conclusions of the relevant studies listed above would be all invalid because they are only based on the researchers’ own particular intuitions, and thus could not be taken to be representative of Korean in general. Furthermore, given the existing binding theories in the generative tradition (e.g., Chomsky 1981; Reinhart 1983, 1986; Grodzinsky & Reinhart 1993), that one pronominal element exhibits two contradictory binding possibilities must be considerably challenging to account for.4 The overarching goal of this chapter is to establish a solid and extensive empirical base for obtaining a complete and precise picture of the interpretative status of Korean (third- person) pronouns. To meet this objective, three experimental studies (Experiments 1A, 1B, and 2) were carried out. It was found that some speakers of Korean consistently accepted the quantificational binding of the pronoun ku while other speakers of Korean consistently rejected it, thus substantiating an existence of inter-speaker variation in the bound variable construal for ku. Another primary goal of this chapter is to provide a principled account for why and how the phenomenon of inter-speaker variation arises and exists in Korean and how it can be captured in binding-theoretic terms. Taking into consideration the historical background of ku and its present status, I have come to the conclusion that child learners of Korean may not receive sufficient and clear evidence regarding the grammar of ku from the primary language input data. Following Han, Lidz, and Musolino (2007), I propose that due to such a deficiency in input, the child learners must randomly “choose” between two competing grammars of ku, only one of which allows a bound variable construal, and this thus results in two different groups of speakers in the Korean speech community. I then attempt to provide theoretical support for this line of analysis by adopting the pronominal typology in Déchaine and Wiltschko (2002), wherein each type of pronoun has a distinct binding-theoretic status.

2.2 Theoretical Background

In this section, I review the existing treatments of the availability of pronominal binding in a variety of languages, which may be divided into three groups: competition-based approaches, constraint-based approaches, and structural approaches. I then attempt to evaluate whether and how they can be adapted to elucidate the status of Korean pronouns.

4Following Tanya Reinhart’s theory of anaphora, I assume that there are two types of pronominal anaphora, variable binding and coreference, and that the Binding Conditions (i.e., the grammar) only governs variable binding.

19 2.2.1 Competition-based Approach

It has been widely reported in the literature that the so-called pro-drop languages generally do not allow overt pronouns to have a bound variable interpretation. Consider the following sets of sentences in various pro-drop languages.

(6) Farsi

a. Che kasi1 goft ke pro1 yek ketab kharid? what one say.past comp a book buy.past

‘Who1 said that he1 bought a book?’

b.* Che kasi1 goft ke oo1 yek ketab kharid? what one say.past comp he a book buy.past

‘Who1 said that he1 bought a book?’ (Tayyebi et al. 2011: 222, ex.(1a) and (1c))

(7) Spanish

a. Nadie1 cree que pro1 es inteligente. nobody believe.pres comp be.pres intelligent

‘Nobody1 believes that he1 is intelligent.’

b.* Nadie 1 cree que él1 es inteligente. nobody believe.pres comp he be.pres intelligent

‘Nobody1 believes that he1 is intelligent.’ (Montalbetti 1984: 93, ex.(33a) and (33b))

(8) Italian

a. Ogni studente1 crede che pro1 è intelligente. every student think.pres comp be.pres intelligent

‘Every student1 thinks that he1 is intelligent.’

b.* Ogni studente1 crede che lui1 è intelligente. every student think.pres comp he be.pres intelligent.

‘Every student1 thinks that he1 is intelligent.’ (Carminati 2013: 3, ex.(1a) and (1b))

(9) Chinese

a. Meige ren1 xiwang pro1 neng xingfu. every man wish can happy

‘Everybody1 wishes that he1 can be happy.’

b.* Meige ren1 xiwang ta1 neng xingfu. every man wish he can happy

‘Everybody1 wishes that he1 can be happy.’ (Xu 1986: 87, ex.(44) and (46))

20 (10) Japanese

a. Dono gakusei1-mo pro1 katu to omotte-iru. every student-part win comp think-pres

‘Every student1 thinks that he1 will win.’

b.* Dono gakusei1-mo kare1-ga katu to omotte-iru. every student-part he-nom win comp think-pres

‘Every student1 thinks that he1 will win.’ (Yashima 2015: 1422, ex.(1) and (4))

In the (a) examples, the null pronouns (pro) in the embedded subject position can take Wh- phrase or quantiﬁer antecedents while the corresponding overt pronouns in the (b) examples cannot.5 According to the competition-based approach, this null/overt asymmetry can be accounted for by postulating that overt pronouns cannot be construed as a bound variable if their “competitors”, null pronouns, can be so construed in the same syntactic position (e.g., Montalbetti 1984; Kanno 1997; Pérez-Leroux & Glass 1999; White 2003).6 Consider now the following sets of sentences in non-pro-drop languages.

(11) English a.* Every student thinks that pro is intelligent.

b. Every student1 thinks that he1 is intelligent.

(12) German a.* Jeder Schuler denkt dass pro intelligent ist. every student think.pres comp intelligent be.pres

b. Jeder Schuler1 denkt dass er1 intelligent ist. every student think.pres comp he intelligent be.pres

‘Every student1 thinks that he1 is intelligent.’

(13) French a.* Personne pense que pro est intelligent. nobody think.pres comp be.pres intelligent 5Note that the overt pronouns in (6)-(10) can all induce a deictic interpretation. 6The competition-based approach was originally put forth by Montalbetti (1984: 94) with the name ‘Overt Pronoun Constraint’.

(i) Overt Pronoun Constraint: Overt pronouns cannot link to formal variables if and only if the alternation empty/overt obtains.

Note that Montalbetti (1984: 48) deﬁnes a ‘formal variable’ as follows.

(ii) Formal variable x is a formal variable iﬀ (i) x is an empty category in an argument position; and (ii) x is linked to a lexical operator in a non-argument position [(e.g., Wh-traces or traces of quantiﬁer raising)].

21 b. Personne pense qu’il est intelligent. nobody think.pres comp.he be.pres intelligent

‘Nobody1 thinks that he1 is intelligent.’ (Fonseca-Greber 2002: 774, ex.(8a) and (8b))

In contrast to their equivalents in pro-drop languages, the overt pronouns in English, Ger- man, and French can all involve quantiﬁcational binding, as shown in the (b) examples above. This can also be easily captured by the competition-based approach because the grammar of these languages does not license null pronominal subjects in embedded clauses, as illustrated in the (a) examples above, and thus the corresponding overt pronouns would “never” compete with any alternative to obtain a bound variable interpretation. A fundamental problem with the competition-based approach, however, is that it does not always yield correct predictions about the binding behaviours of overt pronouns in pro-drop languages.7 In Shuswap, for example, not only a null pronoun but also an overt pronoun can be used to obtain a bound variable interpretation, as illustrated below.8

(14) Shuswap

a. Xwexweyt1 re swet xwis-t-∅-es pro1 re qe7tse-s. all det who like-tr-3sg.obj-3sub det father-3sg.poss

‘Everyone1 likes his1 own father.’

b. Xwexweyt1 re swet xwis-t-∅-es newi7-s1 re all det who like-tr-3sg.obj-3sub emph-3sg det qe7tse-s. father-3sg.poss

‘Everyone1 likes HIS1 own father.’ (adapted from Lai 1998: 31, ex.(21a) and (21b))

Under the competition-based approach, it would be predicted that the overt pronoun newi7- s should not take a quantiﬁed antecedent, contrary to (14b), because the overt pronoun would “lose” to the null pronoun in the competition for a bound variable interpretation. In fact, given that Korean is also a pro-drop language, the two conﬂicting intuitions on the binding behaviour of ku would also be a counterexample to the notion of competition, since

7Note that the Japanese long-distance anaphor zibun ‘self’ can also be interpreted as a bound variable in the context of (10), as shown in (i).

(i) Daremo1-ga zibun1-ga atama-ga ii to omotte-iru. everyone-nom self-nom head-nom good comp think-pres

‘Everyone1 thinks that he1 is intelligent.’

As pointed out by Gürel (2006: 263), the fact that “[zibun] can pattern with the null pronoun in bound variable contexts” would pose a problem to the competition-based approach because it would not be easy to account for why only kare, but not zibun, should compete with and lose out to the null pronoun for a bound variable interpretation. 8Shuswap (also known as Secwepemctsin) is a Northern Interior Salish language spoken in the interior of British Columbia, Canada.

22 ku would be uniformly predicted not to be construed as a bound variable. Consider now the following Japanese sentences.

(15) a.* Daremo 1-ga Mary-ga pro1 sitteiru-to it-ta. everyone-nom Mary-nom know-comp say-past

‘Everyone1 said that Mary knew him1.’

b.* Daremo 1-ga Mary-ga kare1-o sitteiru-to it-ta. everyone-nom Mary-nom he-acc know-comp say-past

‘Everyone1 said that Mary knew him1.’ (Noguchi 1997: 774, ex.(15))

Recall from the examples of non-pro-drop languages, (11)−(13), that the competition-based approach would predict an overt pronoun to be bindable in a syntactic context where a competition with a null pronoun would never hold.9 Such a prediction is not always borne out, however. In Japanese, for instance, a null pronoun cannot be construed as a bound variable in the embedded object position, as illustrated in (15a), but still the overt pronoun kare ‘he’ cannot be used as a bound variable in the same position (Noguchi 1997), as illustrated in (15b).10 Given the discussions so far, it can be concluded that the notion of competition cannot capture the precise distribution and interpretation of overt pronouns in pro-drop languages.

2.2.2 Constraint-based Approach

Several constraint-based explanations have been given in the literature to account for the data of pro-drop languages discussed above. For instance, syntactic constraints such as

9In Spanish (and probably many other pro-drop languages), a null pronoun is not licensed in the prepo- sitional object position (e.g., Montalbetti 1984; Pérez-Leroux & Glass 1999), as the ungrammaticality of (i) indicates. Under the competition-based approach, this constitutes a context where no null vs. overt competition would exist, and it would be thus predicted that the bound variable use of an overt pronoun is legitimate in this context. This prediction seems to be borne out, as illustrated in (ii).

(i)* Todo el mundo dice que el presidente habla de pro. everyone says comp the president speaks of him

(ii) Todo el mundo1 dice que el presidente habla de él1. everyone says comp the president speaks of him

‘Everyone1 says that the president speaks of him1.’ (Pérez-Leroux & Glass 1999: 227, ex.(13), originally from Campos 2002)

10One might attempt to argue that in (15), both the null pronoun and kare ‘he’ cannot be interpreted as bound by the quantiﬁed matrix subject, since zibun ‘self’ can be so interpreted, as shown in (i).

(i) Daremo1-ga Mary-ga zibun1-o sitteiru-to it-ta. everyone-nom Mary-nom he-acc know-comp say-past

‘Everyone1 said that Mary knew him1.’ This line of argument, however, fails to explain examples like (10), where the null pronoun as well as zibun can receive a bound variable interpretation. See M. Y. Kang (1988: 195) for a similar argument.

23 those in (16) have been put forward for the widely-documented observation that Japanese third-person pronouns cannot be bound (but can only refer), as in (17) (repeated from (10b)).11

(16) a. Kare must be A-free.¯ (Aoun & Hornstein 1992: 5) b. Kare must be operator-free. (Katada 1991: 307)

(17)* Dono gakusei1-mo kare1-ga katu to omotte-iru. every student-part he-nom win comp think-pres

‘Every student1 thinks that he1 will win.’

Following May (1977, 1985) in assuming that quantiﬁers undergo Quantiﬁer Raising (QR) to a TP-adjoined position, Aoun and Hornstein (1992) and Katada (1991) attribute the ill- formedness of (17) to the fact that kare is either A-bound¯ or operator-bound, thus violating the constraint in either (16a) or (16b).12 Hong (1985) also proposes a constraint, as stated in (18), which is intended to account for the contrast between the binding properties of overt pronouns in pro-drop and non-pro- drop languages.

(18) Constraint on Pronominal Binding: a. Overt pronouns must be A-free¯ at LF in pro-drop languages. b. Overt pronouns must be locally A-free¯ at LF in non-pro-drop languages. (adapted from Hong 1985: 95)

(19) local binding: X locally binds Y if X binds Y and there is no Z such that X binds Z and Z binds Y. (Chomsky 1984/1986: 164-165)

Adopting Chomsky’s (1984/1986) deﬁnition of local binding given in (19), Hong maintains that overt pronouns in pro-drop languages cannot be A-bound¯ at all; on the other hand, overt pronouns in non-pro-drop languages can be A-bound¯ as long as they are non-locally A-bound.¯ This cross-linguistic diﬀerence is illustrated in (20a) and (20b), where the LF representation for the Japanese sentence in (17) cannot be assigned a bound variable interpretation, but the LF representation for the corresponding English sentence can.13

11Contrary to the “standard” observation, however, it has been reported by some researchers (e.g., Hoji et al. 1999; Hara 2002; Yashima 2015) that kare ‘he’ and kanozyo ‘she’ can be construed as bound variables under certain conditions. In footnote 54, I present Yashima’s (2015) analysis of the interpretative status of Japanese third-person pronouns, including the above seemingly contradictory observations, and discuss how it can theoretically support the proposal of this chapter regarding the inconsistent binding properties of Korean third-person pronouns. 12Katada claims that “A-positions occupied at LF imply operators in the necessary and suﬃcient sense” (p.307), and thus (16b) is another way of saying (16a). 13Hong adopts Chomsky’s (1981: 330) deﬁnition of ‘variable’, as stated in (i).

24 (20) a.* Dono gakuse 1-mo [t1 [kare1-ga katu to] omotte-iru.]

b. Every student1 [t1 thinks [that he1 will win.]]

Under Hong’s approach, the LF in (20a) is not legitimate in that the embedded subject kare is A-bound¯ (i.e., is not A-free),¯ thus violating the constraint in (18a).14 The LF in (20b), on the other hand, is taken to be legitimate in that the embedded subject he is not locally A-bound¯ due to another binder, the trace of the QR’d matrix subject (i.e., is locally A-free),¯ thus abiding by the constraint in (18b). As pointed out by Noguchi (1997: 772), however, an essential problem with the above constraint-based treatments is that a proposed constraint is merely a circular description of the given phenomenon that invites a circular question, and is thus “a restatement of the problem rather than a solution thereof”.15 Under Aoun and Hornstein’s (1992) proposal, for instance, the unavailability of a bound variable construal for kare is attributed to the constraint that kare must be A-free,¯ but it then raises another question: why must kare be A-free?¯ Similarly, with regard to the issue of accounting for the inconsistent binding property of the Korean pronoun ku, it would not be satisfactory to simply suggest a “novel” constraint such as “ku may or may not be A-free”,¯ which would again bring up the question as to why it should be so.

2.2.3 Structural Approach

In this subsection, I brieﬂy review three structural approaches to the diﬀerent binding properties of pronouns in various languages, which have been proposed by Noguchi (1997), Panagiotidis (2002), and Déchaine and Wiltschko (2002), respectively.

(i) α is a variable if and only if it is locally A-bound¯ and is in an A-position.

Notice that the traces left by QR and the pronouns in (20) are variables by deﬁnition. 14Similarly, Hong argues that the Korean pronoun ku cannot be construed as a bound variable, as illustrated in (i).

(i)* Nwukwuna 1 [t1 [ku1-ka toktokha-ta-ko] sanggakha]-n-ta. everyone he-nom intelligent-decl-comp think-pres-decl

‘Everyone1 thinks that he1 is intelligent.’ (Hong 1985: 96, ex.(50b)) Note also that Hong uses the constraints in (18) to explain the standard weak crossover (WCO) cases, as shown in the English and Korean examples in (ii) and (iii).

(ii)* His1 mother loves everyone1.

LF: Everyone1 ... his1 ...... t1

(iii)* Ku1-uy umma-ka nwukwuna1-lul sarangha-n-ta. he-gen mother-nom everyone-acc love-pres-decl

LF: Nwukwuna1 ... ku1 ...... t1

In (ii), his is locally A-bound¯ at LF, and thus violates the constraint in (18a), while in (iii), ku is A-bound¯ at LF, and thus violates the constraint in (18b). 15See also Elbourne (2005) for a similar argument.

25 Following Reinhart (1983, 1986) and Grodzinsky and Reinhart (1993), the three structural approaches share the assumption that there are, at least, two types of anaphoric relations in natural language, (variable) binding and coreference, and that only the former is constrained by the Binding Conditions. They also agree that the cross-linguistic differences in the availability of bound variable construal of pronouns do not derive from some parametric variation within the Binding Theory (e.g., different binding domains), but from different structural characteristics of the pronouns themselves (cf. Déchaine and Wiltschko 2003). However, the fundamental difference between Noguchi (1997) and Panagiotidis (2002) on the one hand and Déchaine and Wiltschko (2002) on the other is whether the grammar (i.e., the Binding Conditions) is sensitive to differences in internal structural organization of pronouns or to differences in their syntactic projections or categories. Consider first (21) from Noguchi (1997: 783) (with irrelevant details suppressed).

(21) a. D-pronoun DP

b. N-pronoun DP

D NP

∅ N

kare

Noguchi proposes that languages can have two types of pronouns: D-pronouns that behave like a determiner and N-pronouns that behave like a lexical noun, which English he and Japanese kare respectively fall into.16 We see in (21a) and (21b) that within the DP structure, he is located at the D node, but kare is located at the N node. Crucially,

16Assuming that lexical items are generally modifiable while functional items are not, Noguchi claims that Japanese pronouns kare ‘he’ and kanozyo ‘she’ are noun-like elements rather than determiner-like elements, citing the fact that just like common nouns, they can be modified by attributive adjectives, as in (i), or by demonstrative determiners, as in (ii) (see also Fukui 1988 for a similar argument). In contrast, English pronouns generally do not take such modifiers, as evident from the unacceptable literal translations in English given below.

(i) a. futota kare fat he (lit.) ‘fat he’

26 assuming the hypothesis that binding can only apply to functional, not lexical, items such as determiners while coreference can apply to both functional and lexical items, Noguchi claims that D-pronouns can be bound as well as be coreferential, but N-pronouns can only be coreferential.17 Panagiotidis (2002: 36) adopts Noguchi’s (1997) pronominal typology, sharing the core assumptions and claims therein, as illustrated in (22).

(22) a. D-pronoun DP

D NumP

he Num NP [def] [sing]

∅ [masc], [pronominal]

b. ryokitekina kanozyo bizarre she (lit.) ‘bizarre she’

(ii) a. kono kare this he (lit.) ‘that he’ b. ano kanozyo that she (lit.) ‘that she’ (Noguchi 1997: 777, originally from Kuroda 1965)

Note also that kare and kanozyo can mean ‘boyfriend’ and ‘girlfriend’, as in the following naturally occurring examples, while their English counterparts he and she cannot.

(iii) a. Watashi-mo kankokujin-no kare-ga imasu. I-also Korean-gen boyfriend-nom exist ‘I also have a Korean boyfriend.’ b. Airi-ga anata-no kanozyo-ni!? Airi-nom you-gen girlfriend-q ‘Airi is your girlfriend!?’ (Google)

According to Noguchi, the above cross-linguistic contrast can be attributed to the lexical/functional distinction since, he argues, only lexical categories can undergo semantic shift. In footnote 55, I present that the Korean counterparts ku and kunye can be modiﬁed by adjectives, but not by demonstrative determiners, and that they show no such semantic changes. I then discuss the implications of these observations for the main proposal of this chapter. 17Noguchi classiﬁes Korean pronouns ku ‘he’ and kunye ‘she’ as D-pronouns, not as N-pronouns, citing M. Y. Kang’s (1988) examples which demonstrate that the pronouns can be construed as bound variables.

27 b. N-pronoun DP

D NumP

∅ Num NP [def] [sing]

kare [masc], [pronominal] (Panagiotidis 2002: 36)

Panagiotidis’ structural approach, however, is distinct from Noguchi’s since it assumes that every pronominal DP has a complex structure (i.e., has a NumP projection), and always contains either a null or overt noun (phrase) that bears a [pronominal] feature, which, he argues, distinguishes pronouns from “ordinary” nouns that carry concept-denoting (or descriptive) features (cf. Postal 1969; Abney 1987). Note that the D-pronoun he in (22a) accommodates a phonologically null NP, and thus is not an intransitive determiner (i.e., is not a determiner with no NP complement), in contrast to Noguchi’s D-pronoun he in (21a). From a purely syntactic perspective, the structural approaches of Noguchi and Panagiotidis can be said to imply that the availability of a bound variable construal for a pronoun is determined by its internal syntactic make-up, i.e., whether it is placed at D or N within a DP projection. Déchaine and Wiltschko (2002), on the other hand, propose that there are (at least) three types of pronouns across languages: Pro-DP, Pro-FP, and Pro-NP.

(23) a. Pro-DP b. Pro-FP c. Pro-NP

DP FP NP

D FP F NP N

tú F NP newí7-s ∅ kare

tl’ò ∅ (Dechaine & Wiltschko 2002: 401)

Assuming that the grammar of binding can only “see” the external category of a pronoun, Déchaine and Wiltschko (2002: 410-411) argue that the three types of pronouns are not

28 uniformly DPs (cf. Postal 1969; Abney 1987), but project to different syntactic categories, DP, FP, and NP, each exhibiting a distinct binding-theoretic property. First, Pro-DPs accommodate a “true DP shell”, and are demonstrably definite. They thus function like R-expressions, subject to Binding Condition C. Pro-FPs, on the other hand, correspond to “any intermediate functional projection that intervenes between D and N” (e.g., NumP), and encode F-features (e.g., number, gender, and person). They play the role of variables, and thus be equivalent to the standard “Condition B pronouns”. Pro-NPs are like lexical nouns, and are inherently semantically constants, undefined regarding its binding-theoretic status.18 Accordingly, in Déchaine and Wiltschko’s theory, only Pro-FPs (e.g., Shuswap newí7-s ‘he or she’ and English he) can be interpreted as a bound variable while Pro-DPs (e.g., (Upriver) Halkomelem tú-tl’ò (det-3sg) ‘he’ and thú-tl’ò (det.fem-3sg) ‘she’) and Pro-NPs (e.g., Japanese kare ‘he’) cannot.19 It is noteworthy that the so-called independent (emphatic) pronouns in Halkomelem and Shuswap, both of which belong to the Salish language family, exhibit contrasting binding properties, as shown in (24) and (25).20

(24) Halkomelem w T w T *M@k’ t @ swaw’l@s1 niĳ ì@q’ @n@t T@ sqeĳ@ts t @-w’-niì1 all det boy aux strike-tr det younger.sibling-3gen det-link-3sg

‘Every boy1 hit HIS1 younger sister.’

(25) shuswap

Xwexweyt1 re swet xwis-t-∅-es newi7-s1 re qe7tse-s. all det who like-tr-3sg.obj-3sub emph-3sg det father-3sg.poss

‘Everyone1 likes HIS1 father.’ (repeated from (14))

Crucially, in contrast to Noguchi (1997), but similar to Panagiotidis (2002), Déchaine and Wiltschko (2002) claim that pronouns always maintain their non-maximal projections; that is, Pro-DPs always contain a FP and an NP, and Pro-FPs an NP. Moreover, the NPs can be phonologically null, as shown in (23a) and (23b) (marked with ‘∅’), “resulting in the ‘pronominal’ use of the pronoun[s]”(p.412).

18Déchaine and Wiltschko (p.411) propose that “[t]he categorical status of these pronominal categories determines their external syntax and their inherent semantics, which in turn determines their binding- theoretic status”. This is illustrated in the following table.

(i) Nominal proform typology Category Pro-DP Pro-FP Pro-NP Internal syntax D syntax neither D or N syntax N syntax Distribution argument argument or predicate predicate Semantics deﬁnite — constant Binding-theoretic status R-expression variable —

19Halkomelem is a Central Coast Salish language spoken in an area of southwestern British Columbia, Canada, and has three principal dialects: Upriver, Downriver, and Island Halkomelem. 20(24) is an Island Halkomelem example from my course project paper for LING 811 (First Nations Language), Simon Fraser University, in 2012.

29 Before ending this section, the point that must be underscored is that in all the three structural approaches discussed so far, the availability of bound variable construal for pronouns is accounted for by the intrinsic essential nature of the pronouns themselves, rather than by the “external” factors such as competition with their alternatives (competition- based approaches) or the nature of their antecedents (constraint-based approaches). Con- sidering the earlier discussion on the problems with the two non-structural approaches, the structural approaches seem to provide a more principled explanation on the issue in question. However, the reported contrasting binding possibilities of the Korean pronoun ku is still diﬃcult for the structural approaches to deal with, since these approaches, in principle, assume that the binding property of a given pronoun correlates (in a one-to-one fashion) with the type it belongs to. If there indeed exists inter-speaker variation regarding the bound variable use of ku, a possible approach compatible with structural approaches would be to postulate that a single pronoun may fall under two distinctive structural types. In Section 2.5, I will adopt this line of approach and attempt to capture the aforemen- tioned variation of ku by postulating two diﬀerent types of ku, each with a distinct binding property.

2.3 Experiment 1

As noted in the outset of this chapter, much disagreement exists in the literature as to the availability of a bound variable construal for ku, but there appears to be a lack of empirical research that have attempted to illuminate the exact nature of ku. Therefore, two truth-value judgment (TVJ) task experiments (Experiment 1A and Experiment 1B) were conducted, and it was found that while some speakers of Korean consistently allowed ku to take a quantified antecedent, other speakers of Korean consistently did not.21 This empirical finding strongly suggests that there is robust inter-speaker variation regarding the bindability of ku. Before moving on to discuss the methods and findings of Experiments 1A and 1B, it should be emphasized that in contrast to the pronoun ku, the long-distance anaphor caki ‘self’ and the null pronoun can readily yield a bound variable interpretation, as illustrated in (26a) and (26b), indicating that native speakers of Korean do have knowledge of variable binding.

(26) a. Motwu1-ka {caki1-uy/pro1} emeni-lul salangha-n-ta. everyone-nom self-gen mother-acc love-pres-decl

21The TVJ task is an experimental methodology for investigating whether a certain interpretation assigned to a sentence can be licensed by a participant’s grammatical competence (Crain & McKee 1985; Crain & Thornton 1998; Gordon 1998). In a typical TVJ task, a discourse context is provided in a story acted out in front of a participant usually using toys, and the participant is asked to judge whether the target sentence is true in the provided context. The TVJ task was originally designed to assess children’s linguistic competence. However, since this method has several advantages, such as reducing performance variables (Gordon 1998), it has also been popularly adopted in the literature to investigate adults’ grammars.

30 ‘Everyone1 loves his1 mother.’

b. Motwu1-ka {caki1-ka/pro1} chwukkwu-lul cal ha-n-ta-ko everyone-nom self-nom soccer-acc well do-pres-comp-decl sayngkakha-n-ta. think-pres-decl

‘Everyone1 thinks that he1 plays soccer well.’

Note also that it is commonly accepted that ku and the null pronoun can receive an (intrasentential) coreferential reading, as illustrated in (27a) and (27b).22,23

(27) a. Minswu1-ka {ku1-uy/pro1} emeni-lul salangha-n-ta. Minswu-nom he-gen mother-acc love-pres-decl

‘Minswu1 loves his1 mother.’

b. Minswu1-ka {ku1-ka/pro1} chwukkwu-lul cal ha-n-ta-ko Minswu-nom he-nom soccer-acc well do-pres-decl-comp sayngkakha-n-ta. think-pres-decl

‘Minswu1 thinks that he1 plays soccer well.’

2.3.1 Experiment 1A

Experiment 1A was conducted to address the following research question.24

(28) research question: Can the Korean pronoun ku be construed as a bound variable? (Is there inter-speaker variation regarding the bound variable use of ku?) 22Caki ‘self’ is also available at the position of ku and the null pronoun in (27a) and (27b), and is construed as a bound variable (e.g., Cho 1996; Kim & Yoon 2009; Han & Storoshenko 2012).

(i) Minswu1-ka caki1-uy emeni-lul salangha-n-ta. Minswu-nom self-gen mother-acc love-pres-decl

‘Minswu1 loves his1 mother.’

(ii) Minswu1-ka caki1-ka chwukkwu-lul cal ha-n-ta-ko sayngkakha-n-ta. Minswu-nom self-nom soccer-acc well do-pres-decl-comp think-pres-decl

‘Minswu1 thinks that he1 plays soccer well.’

23It is a generally accepted claim that a pronoun can be anaphorically related to a referential antecedent via binding as well as coreference. Therefore, one might reasonably say that the attested readings with ku in (27) could have been obtained through the binding mechanism. This issue will be touched upon in Chapters 3 and 4, instead of in the current chapter. 24The work based on Experiment 1A was presented as a poster at the 22nd Japanese/Korean Linguistics Conference at NINJAL, Japan, in October 2012.

31 2.3.1.1 Methodology

2.3.1.1.1 Participants

Eighteen adult native Korean speakers living in Vancouver, Canada, participated in Ex- periment 1A.25 Socio-linguistic factors such as gender, age, and social status were assumed to not be associated with bound variable anaphora, and thus were not controlled in this experiment. In order to minimize the probable L2 inﬂuence on L1 judgment, however, participants were required to have spent no more than 5 months living in countries other than Korea and to have never attended any non-Korean schools before the age of 16. The participants were aged between 19 and 29 (the mean age was 23). Most of them were ESL students, and the rest were visitors with working holiday visas.

2.3.1.1.2 Task

A TVJ task was employed in Experiment 1A. First, sentences describing a context followed by a target sentence were presented to participants on a computer screen. The participants were then asked to judge if the target sentence truthfully described the given context by entering 1 for ‘True’ and 0 for ‘False’ (see Figure 2.1 below).

한수, 진수, 민수가 농구장에서 농구를 하고 있었다.

한수는 자기가 농구를 잘 한다고 말했다. 진수도 자기가 농구를 잘 한다고 말했다. 민수도 자기가 농구를 잘 한다고 말했다.

모두가 농구장에서 그가 농구를 잘 한다고 말했다.

Figure 2.1: Screenshot of a test trial in Experiment 1A

25All experimental sessions were conducted in a classroom on the Burnaby campus of Simon Fraser Uni- versity.

32 Antecedent Type Pronoun Type Quantiﬁcational Overt Null Referential Overt Null

Table 2.1: Test conditions in Experiment 1A

2.3.1.1.3 Design and Materials

Each target sentence was a complex sentence containing an embedded clause, where the matrix subject was either motwu ‘everyone’ or a proper name, and the embedded subject was ku or a null pronoun. Each context was biased towards an interpretation of the embedded subject anaphorically linked to the matrix subject. Hence, two within-subjects factors were crossed to create four conditions: antecedent type (Quantiﬁcational vs. Referential) × pronoun type (Overt vs. Null). Table 2.1 above summarizes the experimental design. Consider now a sample set of test items, given in (29)-(32), where the target sentences are in boldface.26

(29) Quantificational-Overt condition:

Hanswu, Cinswu, Minswu-ka nongkwucang-eyse nongkwu-lul ha-ko Hanswu, Cinswu, Minswu-nom basketball.court-at basketball-acc do-prog iss-ess-ta. Hanswu-ka caki-ka nongkwu-lul cal ha-n-tako past-decl Hanswu-nom self-nom basketball-acc well do-pres-comp malha-yess-ta. Cinswu-to caki-ka nongkwu-lul cal ha-n-tako say-past-decl Cinswu-also self-nom basketball-acc well do-pres-comp malha-yess-ta. Minswu-to caki-ka nongkwu-lul cal ha-n-tako say-past-decl Minswu-also self-nom basketball-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu said that he plays basketball well. Cinswu also said that he plays basketball well. Minswu also said that he plays basketball well.’

Motwu-ka nongkwucang-eyse ku-ka nongkwu-lul cal everyone-nom basketball.court-at he-nom basketball-acc well ha-n-tako malha-yess-ta. do-pres-comp say-past-decl

26A complete set of all test items can be found in Appendix A. In Experiment 1A and the other two experiments (1B and 2), each condition appeared four times. In each instance, the background scenes, characters, places, and topics were diﬀerent. For the names of characters, typical Korean male names were used.

33 ‘Everyone said at a basketball court that he plays basketball well.’

(30) Quantificational-Null condition: (context sentences are the same as those in (29))

Motwu-ka nongkwucang-eyse pro nongkwu-lul cal everyone-nom basketball.court-at basketball-acc well ha-n-tako malha-yess-ta. do-pres-comp say-past-decl ‘Everyone said at a basketball court that he plays basketball well.’

(31) Referential-Overt condition:

Hanswu-ka nongkwucang-eyse ku-ka nongkwu-lul cal ha-n-tako Hanswu-nom basketball.court-at he-nom basketball-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Hanswu said at a basketball court that he plays basketball well.’

(32) Referential-Null condition: (context sentences are the same as those in (31))

Hanswu-ka nongkwucang-eyse pro nongkwu-lul cal ha-n-tako Hanswu-nom basketball.court-at basketball-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Hanswu said at a basketball court that he plays basketball well.’

In (29) and (30), each context is compatible with the quantificational binding interpretation for either ku or the null pronoun in the target sentence. In (31) and (32), on the other hand, each context is compatible with the coreferential interpretation for either ku or the null pronoun in the target sentence. The condition that was key to the experiment was the Quantificational-Overt condition, since the primary concern of the present study is whether ku can be interpreted as a bound variable. The Referential-Overt, Referential-Null, and Quantificational-Null conditions were treated as control (or comparison) conditions, where

34 the intended readings of the target sentences were expected to be highly accepted, given that ku can be coreferential and a null pronoun can be coreferential or bound, as discussed previously. It should also be noted that the experimental design made it possible to additionally examine the tenability of the competition-based approach discussed in Section 2, through a comparison of the participants’ responses in the Quantificational-Overt and Quantificational-Null conditions. Filler items were also constructed, which were similar in structure to the test items. (33) and (34) are examples of the fillers, in which the matrix subject of each target sentence is either motwu ‘everyone’ or a proper name, and each context is compatible with the ‘discourse binding’ interpretation for ku in the target sentence. These items were also taken to be controls, where high acceptance rates were expected to be obtained since ku can readily take an extra-sentential antecedent, as discussed in footnote 3.

(33) Quantificational-discourse condition:

Hanswu, Cinswu, Minswu-ka nongkwucang-eyse, Kiswu-ka oki-lul Hanswu, Cinswu, Minswu-nom basketball.court-at Kiswu-nom coming-acc kitalimye, nongkwu-lul ha-ko iss-ess-ta. Hanswu-ka Kiswu-ka waiting basketball-acc do-prog past-decl Hanswu-nom Kiswu-nom nongkwu-lul cal ha-n-tako malha-yess-ta. Cinswu-to Kiswu-ka basketball-acc well do-pres-comp say-past-decl Cinswu-also Kiswu-nom nongkwu-lul cal ha-n-tako malha-yess-ta. Minswu-to Kiswu-ka basketball-acc well do-pres-comp say-past-decl Minswu-also Kiswu-nom nongkwu-lul cal ha-n-tako malha-yess-ta. basketball-acc well do-pres-comp say-past-decl ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court, waiting for Kiswu to come. Hanswu said that Kiswu plays basketball well. Cinswu also said that Kiswu plays basketball well. Minswu also said that Kiswu plays basketball well.’

Motwu-ka nongkwucang-eyse ku-ka nongkwu-lul cal everyone-nom basketball.court-at he-nom basketball-acc well ha-n-tako malha-yess-ta. do-pres-comp say-past-decl ‘Everyone said at a basketball court that he plays basketball well.’

(34) Referential-discourse condition:

35 ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court, waiting for Kiswu to come. Hanswu said that Kiswu plays basketball well.’

Hanswu-ka nongkwucang-eyse ku-ka nongkwu-lul cal ha-n-tako Hanswu-nom basketball.court-at he-nom basketball-acc well do-pres-comp malha-yess-ta. say-past-decl

‘Hanswu said at a basketball court that he plays basketball well.’

2.3.1.1.4 Procedure

The experiment was implemented using the WebExp software package (Keller et al. 2009), and was run on a personal computer. The experimental session began with three practice trials to train the participants to fully understand their task. After the practice trials, 16 test trials (i.e., four trials per condition) and 40 ﬁller trials followed in a uniquely generated random order. On average, the participants took 10-15 minutes to complete the entire experiment. They were paid $5 each as compensation for participation.

2.3.1.2 Findings and Discussion

Figure 2.2 below summarizes mean rates of acceptance (i.e., assignment of 1 ‘True’) by condition: 54% in the Quantificational-Overt condition, 96% in the Quantificational-Null condition, 83% in the Referential-Overt condition, and 92% in the Referential-Null condition. Data were fitted to generalized linear mixed-effects models using the ‘lmer’ function of the ‘lme4’ package (Bates et al. 2012) in the R statistical software (R Development Core Team 2012) to analyze the participants’ responses as a function of antecedent type and pronoun type, with participants and items as random effects. First, the analysis revealed a main effect of pronoun type (coefficient estimate = −4.18, se = .77, z = −5.43, p < .001): regardless of antecedent type, speakers were significantly more likely to accept sentences with null pronouns than overt pronouns (i.e., ku). Second, a significant interaction between pronoun type and antecedent type was found (coefficient estimate = 2.87, se = .95, z = 3.02, p < .01): for overt pronouns, speakers were more likely to accept referential subjects (i.e., proper names) as antecedents than quantificational subjects (i.e., motwu ‘everyone’), while equally likely to accept both quantificational and referential subjects as antecedents for null pronouns. Pairwise comparisons of the mean acceptance rates using Tukey’s tests revealed that in the quantificational subject conditions, null pronouns were significantly more likely to be accepted than overt pronouns (p < .001) while in the referential subject conditions, the acceptance rates of overt and null pronouns are not significantly different from each other. The analysis also revealed that for overt pronouns, referential subjects were significantly more likely to be accepted as antecedents than quantificational subjects (p < .001), but for

36 Overt Null 100 80 60 40 Acceptance rate(%) Acceptance 20 0 Quantificational Referential

Figure 2.2: Mean rates of acceptance and standard errors in Experiment 1A null pronouns, no significant difference was found between the acceptance rates of quantificational and referential subjects. Taken together, the results suggest that the acceptance rates in the Referential-Overt, Referential-Null, and Quantificational-Null conditions (but not the Quantificational-Overt condition) are uniformly high. In fact, this is predicted from and compatible with the earlier discussion on coreference and bound variable anaphora in Korean: ku can enter into coreference, and a null pronoun can enter into both coreference and binding. This result strongly supports that the experimental design was appropriate to assess participants’ grammars of anaphora, thus ensuring that the 54% acceptance rate in the Quantificational-Overt condition, the main concern of the experiment, was indeed a result of ku being interpreted through the variable binding mechanism.27 In order to characterize the acceptance rate in the Quantificational-Overt condition, the data from each participant was further inspected in the following way. Following the common practice in the literature (e.g., Han et al. 2007; J. S. Kim 2012), all participants (n=18) were assigned into three different groups according to their mean individual acceptance rates on the bound variable interpretation for ku: accept (> 75% acceptance: assignment of 1 to three or four target sentences), ambivalent (= 50% acceptance: as-

27The mean acceptance rates in the Quantificational- and Referential-Discourse filler conditions ((33) and (34)) were 94% and 92%, both of which were not significantly different from all the test conditions except for the Quantificational-Overt condition. These results are compatible with the widely held view that ku can readily induce a deictic reading, further ensuring the reliability of the results obtained in the test conditions.

37 10 8 6 4 Number of participants Number 2 0 Reject (0-25%) Ambivalent (50%) Accept (75-100%)

Figure 2.3: Distribution of responses in Quantificational-Overt condition in Experiment 1A signment of 1 to two target sentences), and reject (< 25% acceptance: assignment of 1 to none or one target sentence). Consequently, a bimodal distribution of responses was observed, as demonstrated in Figure 2.3 above: the participants either nearly always accepted or nearly always rejected the bound variable interpretation for ku. In other words, the 54% in the Quantificational-Overt condition is not derived from each participant accepting the bound variable reading of ku 54% of the time, but from about half of the participants who consistently accepted it. This finding supports the presence of between-speaker, not within-speaker, variation in Korean regarding the binding property of ku.28,29 As mentioned previously, Experiment 1A was also designed to investigate whether there is a competition between overt and null pronouns for a bound variable interpretation. Re-

28As pointed out by William O’Grady (personal communication, May 2016), a discourse antecedent is generally preferred over an intra-sentential antecedent for ku (see also Suh 1990 and H. N. Kim 2007). However, the high mean acceptance rate in the Referential-Overt condition in Figure 2.2 indicates that ku can have a (referential) antecedent in the same sentence as long as an appropriate context is provided. This thus implies that the result obtained in the Quantificational-Overt condition is not relevant to the “general” dispreference of ku to take an intra-sentential antecedent. 29One might possibly maintain that the split among the participants in the availability of the bound variable reading for ku could have been attributed to their inability to “entertain” the other interpretation once one of the two possible interpretations (i.e., a bound variable reading and a referential reading) was initially accessed. However, this line of argument cannot be sustained since it fails to provide a principled account for why then such a split was never found in the other test and filler conditions, e.g., Quantificational- Null, Referential-Overt, Referential-Null, Quantificational-Discourse and Referential-Discourse conditions, where one of two possible interpretations would have been initially accessed to the participants in the same way as in the Quantificational-Bound condition, but high acceptance rates were uniformly observed.

38 call that according to the competition-based approach (e.g., Montalbetti 1984), an overt pronoun cannot be construed as a bound variable if a null pronoun can be so construed in the same configuration. Given that the acceptance rate in the Quantificational-Null condition is near 100%, the 54% acceptance in the Quantificational-Overt condition viti- ates the validity of the competition-based approach in a strict sense, as it would predict (near) 0% acceptance of bound variable interpretation for ku.30 Additionally, under the competition-based approach, a split found between speakers regarding the acceptance of bound variable ku would mean that there are two groups of speakers: those who accepted only the bound variable reading of pro belong to the group who is sensitive to the putative “pro-ku competition mechanism”, and those who accepted the bound variable reading of ku as well as pro belong to the group who is not sensitive to the mechanism at all. This is, however, an undesirable explanation which raises a rather complicated question as to what would cause such differences in speakers’ sensitivity to the competition mechanism. Given the other problems with the competition-based approach discussed earlier, it is difficult to conclude that ku indeed competes with a null pronoun for a bound variable interpretation.

2.3.2 Experiment 1B

Experiment 1B was motivated by M. Y. Kang’s (1988) observation that the availability of a bound variable construal of ku may diﬀer depending on clause type. Kang reports that a bound variable reading of ku is ﬁne in a single clause, as shown in (35a), while it is, according to his informants’ intuitions, very marginal across a clause boundary, as shown in (35b).31 Note that only sentences with embedded clauses, as in (35b), were employed as test stimuli in Experiment 1A.32

(35) a. Motwu1-ka ku1-uy emeni-lul coaha-n-ta. everyone-nom he-gen mother-acc like-pres-decl

‘Everyone1 likes his1 mother.’ (adapted from M. Y. Kang 1988: 195, ex.(37))

b.?? Motwu 1-ka ku1-ka hyenmyengha-ta-ko sayngkakha-n-ta. everyone-nom he-nom wise-comp-decl think-pres-decl

‘Everyone1 thinks that he1 is wise.’ (adapted from M. Y. Kang 1988: 194, ex.(33))

Experiment 1B was thus carried out to address the following research questions.33

30See Kweon (2017) for a similar conclusion reached in her experimental research on the bindability of ku, which uses a forced-choice picture-matching task with audio stimuli, but the same experimental design as in Experiment 1A. 31In Kang’s original data, nwukwuna ‘everyone’ is used instead of motwu. 32Kang maintains that ku generally can be interpreted as a bound variable while the degraded cases such as (35b) are exceptions, and are probably controlled by some unknown pragmatic factors. 33The work based on Experiment 1B was presented at the Impact of Pronominal Form on Interpretation Workshop at the University of Tübingen in November 2013.

39 (36) research questions: (i) Is there a diﬀerence between the bound variable construal of ku in a single clause and across clauses? And subsequently, (ii) Could the ﬁndings from Experiment 1A (i.e., the inter-speaker variation phenomenon) be further substantiated?

2.3.2.1 Methodology

2.3.2.1.1 Participants

Thirty-seven native Korean adult speakers living in Vancouver, Canada, who did not participate in Experiment 1A, participated in Experiment 1B. They all met the same eligibility requirement for participation as in Experiment 1A. They were between the ages of 20 and 26 (the mean age was 24), and were all university students in Korea, but came to Canada to work part-time or study English at ESL institutions temporarily.

2.3.2.1.2 Task

The same TVJ task was employed as in Experiment 1A. That is, the participants were instructed to judge whether a target sentence was a true or false description of a given context by entering 1 for ‘True’ and 0 for ‘False’.

2.3.2.1.3 Design and Materials

Each target sentence was either a simple sentence or a complex sentence containing an embedded clause, both of which had motwu ‘everyone’ as a matrix subject. The simple sentence contained ku as a possessor while the complex sentence contained ku as an embedded subject. Each given context “forced” ku to be interpreted as anaphorically linked to the matrix subject. The experiment thus had one within-subjects factor with two levels: clause type (Simple vs. Complex). Consider a sample set of test items below, where the target sentences are in boldface.34

(37) simple condition: Hanswu, Cinswu, Minswu-ka nongkwucang-eyse nongkwu-lul ha-ko Hanswu, Cinswu, Minswu-nom basketball.court-at basketball-acc do-prog iss-ess-ta. Hanswu-ka caki-uy umlyoswu-lul masi-ess-ta. Cinswu-to past-decl Hanswu-nom self-gen beverage-acc drink-past-decl Cinswu-also caki-uy umlyoswu-lul masi-ess-ta. Minswu-to caki-uy umlyoswu-lul self-gen beverage-acc drink-past-decl Minswu-also self-gen beverage-acc masi-ess-ta. drink-past-decl

34A complete set of all test items can be found in Appendix B. The test items for the Complex condition were drawn from the Quantiﬁcational-Overt condition in Experiment 1A.

40 ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu drank his own beverage. Cinswu also drank his own beverage. Minswu also drank his own beverage.’

Motwu-ka nongkwucang-eyse ku-uy umlyoswu-lul masi-ess-ta. everyone-nom basketball.court-at he-gen beverage-acc drink-past-decl

‘Everyone drank his beverage at a basketball court.’

(38) complex condition: (repeated from (29))

‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu said that he plays basketball well. Cinswu also said that he plays basketball well. Minswu also said that he plays basketball well.’

Motwu-ka nongkwucang-eyse ku-ka nongkwu-lul cal everyone-nom basketball.court-at he-nom basketball-acc well ha-n-tako malha-yess-ta. do-pres-comp say-past-decl

‘Everyone said at a basketball court that he plays basketball well.’

2.3.2.1.4 Procedure

Experiment 1B used the same procedure as in Experiment 1A, except that eight test trials (i.e., four trials per condition) and 46 ﬁller trials were presented to the participants in a uniquely generated random order.35 The participants took 10-15 minutes to complete the entire experiment, and were paid $10 each for their participation.

2.3.2.2 Findings and Discussion

Figure 2.4 below summarizes the mean acceptance rates by condition: 43% in the Simple condition and 36% in the Complex condition. Generalized linear mixed-eﬀects models were

35The ﬁller items were taken from Experiment 1A.

41 Simple Complex 100 80 60 40 Acceptance rate(%) Acceptance 20 0

Figure 2.4: Mean rates of acceptance and standard errors in Experiment 1B used to analyze the participants’ responses as a function of clause type, with participants and items as random effects. The analysis revealed no significant effect of clause type, indicating that the acceptance rates in the Simple and Complex conditions were not significantly different from each other, although the former was numerically higher than the latter. As in Experiment 1A, the acceptance rates of each participant in both conditions were examined. It was found that some participants consistently accepted the bound variable interpretation of ku while others consistently rejected it, as illustrated in Figure 2.5 below. In other words, the 43% and 36% acceptance rates in the Simple and Complex conditions were derived from 43% and 36% of the participants who always accepted the bound variable reading of ku, not from each participant accepting it 43% and 36% of the time. Therefore, as in Experiment 1A, these findings substantiate the existence of inter-speaker variation regarding the binding property of ku. A linear regression analysis was carried out to test the correlation between the participants’ mean acceptance rates in both conditions. The analysis revealed that the correlation coefficient is significantly different from zero (r = .62, t = 7.49, p < .001), indicating that an individual speaker’s acceptance of the bound variable reading in the Complex condition is predictable from her acceptance of the bound variable reading in the Simple condition. In other words, those participants who accepted the bound variable reading of ku did so for both simple and complex clause types, and those who rejected the bound variable reading did so for both clause types (see Figure 2.6 below). Taken together, it can be concluded

42 25 Simple Complex 20 15 10 Number of participants Number 5 0 Reject(0-25%) Ambivalent(50%) Accept (75-100%)

Figure 2.5: Distribution of responses in Simple and Complex conditions in Experiment 1B

Figure 2.6: Correlation between mean acceptance rates in Simple and Complex conditions in Experiment 1B

43 that there is no significant difference between the availability of a bound variable construal of ku in a single clause or across a clause boundary, contradicting M. Y. Kang’s (1988) observation that clause type may affect the bound variable construal of ku.

2.4 Experiment 2

We have concluded from the ﬁndings of Experiments 1A and 1B that there does exist considerable variation across speakers as to whether the Korean pronoun ku can be construed as a bound variable. At this point, recall from footnote 21 that a TVJ task is assumed to be a methodology that can access a participant’s grammatical competence or knowledge. That is, a participant’s judgment in a TVJ task, whether it is acceptance or rejection, can be interpreted as revealing the state of her grammar regarding the linguistic phenomenon being investigated (Crain & McKee 1985; Crain & Thornton 1998; Gordon 1998). In light of this, we can reasonably assume that the phenomenon of inter-speaker variation observed in Experiments 1A and 1B is a reﬂection of speakers each having a distinct grammar of ku. That is, some speakers may have a grammar that does license a bound variable construal for ku, while others may have a grammar that does not license it. To the extent that this assumption is on the right track, it would then be predicted that an individual speaker’s judgment regarding the bindability of ku would be consistent over time, since she would, in principle, maintain one single grammar for ku. Experiment 2 was thus conducted to address the following research question.36

(39) research question: Do Korean speakers exhibit the same judgment over time on the bound variable construal of ku?

2.4.1 Methodology

2.4.1.1 Participants

Thirty-seven native Korean adult speakers living in Vancouver, Canada, who did not participate in any of the previous experiments, participated in Experiment 2. The participants met the same eligibility requirements for participation as in the previous experiments, and were between the ages of 19 and 28 (the mean age was 23). Most of them were university students in Korea, but came to Canada to work part-time or study English at ESL institutions temporarily.

36The study based on Experiment 2 was presented at the joint meeting of the 19th International Circle of Korean Linguistics Conference (ICKL19) and the 16th Harvard International Symposium on KoreanLin- guistics (ISOKL16) at the University of Chicago in July 2015.

44 context type test session Bound August September Free August September

Table 2.2: Test conditions in Experiment 2

2.3.3.1.1 Task

The same TVJ task was used as in the previous experiments.

2.4.1.2 Design and Material

The experiment consisted of two diﬀerent test sessions that took place one month apart (August and September 2014). Each target sentence was a simple sentence containing motwu ‘everyone’ as the subject and ku as the possessor. Each context was biased towards the bound or free (deictic) interpretation of ku in the target sentence. Thus, two within- subjects factors with two levels each were tested: context type (Bound vs. Free) × test session (August vs. September). The test conditions are summarized in Table 2.2 above. Crucially, the test items were diﬀerent between sessions.37 (40) and (41) illustrate a sample set of test items, where the target sentences are in boldface.38

(40) bound condition: (repeated from (37))

Hanswu, Cinswu, Minswu-ka nongkwucang-eyse nongkwu-lul ha-ko Hanswu, Cinswu, Minswu-nom basketball.court-at basketball-acc do-prog iss-ess-ta. Hanswu-ka caki-uy umlyoswu-lul masi-ess-ta. Cinswu-to past-decl Hanswu-nom self-gen beverage-acc drink-past-decl Cinswu-also caki-uy umlyoswu-lul masi-ess-ta. Minswu-to caki-uy umlyoswu-lul self-gen beverage-acc drink-past-decl Minswu-also self-gen beverage-acc masi-ess-ta. drink-past-decl

‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu drank his own beverage. Cinswu also drank his own beverage. Minswu also drank his own beverage.’

37Filler items were taken from Experiments 1A and 1B. Unlike the test items, the ﬁller items were the same between sessions. 38A complete set of all test items can be found in Appendix C. The test items for the Bound-August condition were adapted from those for the Simple condition in Experiment 1B.

45 Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’

(41) free condition:

Hanswu, Cinswu, Minswu-ka nongkwucang-eyse, Kiswu-ka oki-lul Hanswu, Cinswu, Minswu-nom basketball.court-at Kiswu-nom coming-acc kitalimye, nongkwu-lul ha-ko iss-ess-ta. Hanswu-ka Kiswu-uy waiting basketball-acc do-prog past-decl Hanswu-nom Kiswu-gen umlyoswu-lul masi-ess-ta. Cinswu-to Kiswu-uy umlyoswu-lul beverage-acc drink-past-decl Cinswu-also Kiswu-gen beverage-acc masi-ess-ta. Minswu-to Kiswu-uy umlyoswu-lul masi-ess-ta. drink-past-decl Minswu-also Kiswu-gen beverage-acc drink-past-decl ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court, waiting for Kiswu to come. Hanswu drank Kiswu’s beverage. Cinswu also drank Kiswu’s beverage. Minswu also drank Kiswu’s beverage.’

Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’

2.4.1.3 Procedure

The experiment for each session was implemented using PsychoPy software (Peirce 2007, 2009). The same procedure was employed as in the previous experiments, except that eight test items (i.e., four per condition) and 56 ﬁller items were presented in a uniquely generated random order. The participants took 15-25 minutes to complete the entire experiment of each session, and were compensated $20 for participation.

2.4.2 Findings and Discussion

The mean acceptance rates by condition are summarized in Figure 2.7 below. The results for both sessions are numerically similar: 37% and 87% in the Bound-August condition and the Free-August condition, and 39% and 84% in the Bound-September condition and the Free- September condition.39 A bimodal distribution of participants’ responses in the Bound condition was observed in each session, as illustrated in Figure 2.8 below: participants tended to either always accept the bound variable reading of ku or always reject it. Generalized linear mixed-eﬀects models were constructed to analyze the participants’ responses as a function of context type and test session, with participants and items

39As in the Quantiﬁcational-Discourse and Referential-Discourse ﬁller conditions in Experiment 1A (see again footnote 27), the mean acceptance rates in the Free conditions were uniformly high, assuring the accuracy and reliability of the experimental data obtained.

46 Bound Free 100 80 60 40 Acceptance rate(%) Acceptance 20 0 August September

Figure 2.7: Mean rates of acceptance and standard errors in Experiment 2 25 August September 20 15 10 Number of participants Number 5 0 Reject(0-25%) Ambivalent(50%) Accept (75-100%) Figure 2.8: Distribution of responses in Bound conditions for August and September sessions in Experiment 2

47 Figure 2.9: Correlation between mean acceptance rates in Bound conditions for August and September sessions in Experiment 2 included as random effects. The analysis revealed no main effect of test session, and no interaction between context type and test session. Thus, no significant difference was found in the patterns of participants’ responses in the Bound-August and Bound-September conditions: in both sessions, only about 40% of the participants accepted the bound variable reading for ku. Furthermore, a linear regression analysis revealed a strong correlation (r = 0.74, t = 6.43, p <. 001) between the mean acceptance rates in the Bound-August and Bound-September conditions, as illustrated in Figure 2.9 above. This indicates that participants who accepted the quantificational binding of ku in August session did so as well in September session, and those who rejected the quantificational binding of ku in August session did so as well in September session. Taken all together, it can reasonably be concluded that each participant did exhibit the same judgment over time on the availability of a bound variable construal for ku, thus confirming the possibility that the phenomenon of inter-speaker variation may be a consequence of existence of two distinct grammars for ku. We will see in the next section that these findings can be seen as supporting the main proposal of this chapter based on the two grammar hypothesis by Han, Lidz, and Musolino (2007).

48 2.5 General discussions

Using an experimental methodology, novel empirical evidence has been established that there is indeed variation across Korean speakers regarding the bindability of ku, as a synchronically “active” linguistic phenomenon. Therefore, it seems that the lack of consensus on the status of ku in the literature is actually a reflection of this inter-speaker variation. An important question arises at this point: why and how does the variation phenomenon emerge and exist in Korean language? As a preliminary step in attempting to provide a principled answer to this question, we first consider the historical background of ku and its present status, from the perspective of language acquisition. Interestingly enough, the Korean third-person pronoun ku has a somewhat idiosyncratic status among pronouns across languages. That is, in contrast to pronouns in other languages such as he in English, it is a rather rare pronominal item which was “artificially” introduced into the language by a certain group of speakers. It is generally accepted by Korean linguists that ku was first used by Korean writers in the early 20th century (K. M. Lee 1978; H. C. Kim 1981; C. H. Kim 1984; Y. H. Kim 2016b).40 It is worth noting the following extract from the autobiographical essay by the novelist Dong-In Kim, who is generally considered to have pioneered the use of ku as a third-person pronoun.41

“I sometimes felt annoyed and tired of repeating the same names to refer to a character in a novel or an essay. I wanted to use a lexical item corresponding to English he or she, but Korean has nothing like that [...] Among the candidates for a third-person pronoun, the demonstrative ku ‘that’ seemed to be nicer conventionally than the demonstrative ce ‘that’ and others, so I just chose it and used it without any hesitation.” (D. I. Kim 1948, compiled by C. H. Kim 1984: 424)

As implied from above, the pronominal ku historically originated from and is thus ho- mophonous with the demonstrative ku ‘that’, which forms one part of the Korean tripartite demonstrative system, as illustrated in (42) and (43).42 Note that in deictic contexts, ku is used to indicate an entity that is closer to the hearer than the speaker,43 while i ‘this’ is

40Ku ‘he’ was initially used only as a gender-neutral pronoun. Kunye ‘she’ was later coined by those writers who were using ku as a masculine pronoun and thus needed a corresponding feminine form (e.g., An 2008; Y. H. Kim 2016b). 41This quote was originally written in Korean, and translated here into English by the author. 42Koulidobrova and Lillo-Martin (2016: 228) point out that cross-linguistically, demonstrative determiners are often used independently, i.e., without combining with nominals, to function as personal pronouns. “The idea that pronouns are related to demonstratives is not entirely odd. In Eastern Armenian, for instance, demonstratives na, sa, and da are used in lieu of personal pronouns (Kozintseva 1995). The same can be said for the Basque hau’, hori, and hura (Saltarelli et al. 1988) as well as the Korean ku.” We will consider examples of German demonstrative pronouns in (51). 43The demonstrative ku can also be used in anaphoric contexts, and thus can refer to a person or an object that has been mentioned in the previous discourse or that is not invisible but known to both the hearer and

49 used for an entity close to the speaker, and ce ‘that over there’ for an entity distant from both the hearer and the speaker (e.g., Hoji 1990; Sohn 1999).44,45

(42) a. i chayk ‘this book’ b. ku chayk ‘that book’ (or ‘the book’) c. ce chayk ‘that book over there’ (43) a. i salam ‘this person’ the speaker (e.g., Chang 2009; Oh 2010; Ionin et al. 2012). Therefore, ku chayk and ku salam in (42b) and (43b) can alternatively be translated as the book and the person in English. 44The three types of demonstratives can also be combined with a wide variety of bound morphemes, thus forming morphologically “complex” words, as illustrated in the following table.

(i) Demonstrative paradigm in Korean (based on Noguchi 1997 and Chang 2009) i-series ku-series ce-series i-ke(s) ‘this thing’ ku-ke(s) ‘that thing’ ce-ke(s) ‘that thing’ i-len ‘like this’ ku-len ‘like that’ ce-len ‘like that’ i-li ‘this way’ ku-li ‘that way’ ce-li ‘that way over there’ i-ttay ‘this time’ ku-ttay ‘that time’ ce-ttay ‘that way over there’ i-kos ‘here’ ku-kos ‘there’ ce-kos ‘over there’

45Noticeably, the demonstratives i and ce, unlike ku, cannot be used independently as personal pronouns, as shown in (i).

(i) ku/*i/*ce-ka nwukwu-ci? he-nom who-Q ‘Who is he?’ (Yu 2005: 197, ex.(15))

As noted by N. R. Han (2006: 40-41), however, i alone can be used to refer to “an event-like semantic content denoted by a previous clausal segment”, as shown in (ii). (ii) Yosay elini-tul-uy silyek-i nappaci-ko iss-nuntey, i-nun olayn TV recently child-pl-gen sight-nom deteriorate-AuxEnd be-CoordEnd this-top long TV sicheng-uy yenghyang-i-ta. watch-gen inﬂuence-cop-decl ‘Children’s eyesight is deteriorating these days, and this is the inﬂuence of long TV watching.’ (N. R. Han 2006: 41, ex.(48c)) Note also that the plural forms of i and ce, i-tul and ce-tul, can be used as personal pronouns, as shown in (iii) and (iv).

(iii) Taypepwen-un saken kwankyeyca-tul-ey tayha-n phankyel-eyse i-tul-eykey calmos-i supreme.court-top case associate-pl-dat regard-adn sentence-at this-pl-dat guilt-nom eps-tako kyellon-cis-ess-ta. lack-decl conclusion-make-past-decl ‘The Supreme court concluded in its sentence on case associates that there is no guilt to these people.’

(iv) Ce-tul-i wuli cwunim-ul wangwi-eyse kkul-e nayli-ess-tota. that-pl-nom our lord-acc throne-from pull down-past-dec ‘Those people (they) pulled down our lord from the throne.’ (N. R. Han 2006: 41, ex.(48b) and (48d))

50 b. ku salam ‘that person’ (or ‘the person’) c. ce salam ‘that person over there’ (Hoji 1990: 6, ex.(12) and (13))

The use of ku as a personal pronoun seems to have been introduced to Korean speakers about 100 years ago. It is thus a relatively recent linguistic evolution in the Korean language, while the use of ku as a demonstrative determiner has a much longer history (M. Y. Kang 1988). This historical background of ku may have resulted in its unique status in present- day Korean. That is, while the pronoun ku is predominantly used in written texts, it is almost never used in (informal) colloquial speech (O’Grady 1984; S. H. Kang 1990; Suh 1990; Hwang 1991; N. R. Han 2006; An 2008; Oh 2010; B. M. Kim 2016).46 Accordingly, native speakers of Korean are scarcely exposed to the pronominal usage of ku in the early stage of language acquisition.47 Rather, they generally ﬁrst encounter it in written forms in essays, novels, or song lyrics at school age. Even in the written contexts, however, the pronoun ku tends to be avoided or omitted if possible, as shown in (44).48

(44) George-nun (ku-ka) ha-ko-siph-ul ttay-ey (ku-uy) George-top (he-nom) do-AuxEnd-want-RelEnd time-at (he-gen) il-ul ha-n-ta. work-acc do-pres-decl. ‘George does his work when he feels like it.’ (N. R. Han 2006: 38, ex.(47))

Note also that personal names, kinship terms, professional titles (e.g., sensayng-nim ‘(hon- oured) teacher’), and general nouns preceded by demonstratives are frequently preferred over the pronouns ku and kunye in anaphoric contexts (e.g., H. Y. Kim 1989; Y. S. Kang 2004; N. R. Han 2006; Oh 2010; Bak 2015). In (45)-(47), for example, Kimyuna, nwuna ‘sister’, and ku yeca ‘that woman (or the woman)’ appear at positions where the feminine pronoun kunye could otherwise be used (indicated by underline).

(45) Sipil-il, J.Estina-nun pulayntu mywucu-in Kimyuna-uy say hwapo-lul 11th J.Estina-top brand muse-adn Kimyuna-gen new picture-acc kongkayha-yess-ta. Kimyuna-nun i hwapo-eyse ikwukcek-in eymelaltu release-past-decl Kimyuna-top this picture-in exotic-adn emerald kalla-uy cwuelli-lul chakyongha-yess-ta [...] Yelum kwangko sok color-gen jewelry-acc wear-past-decl summer advertisement inside

46Oh (2010: 1222) reports that “there was not even a single instance of [ku and kunye] in the conversational data used in [her] study”. See also H. Y. Kim (1989) and Y. S. Kang (2004) for similar results in their studies on Korean speakers’ choices among overt pronouns (ku and kunye), null pronouns, and full nominal expressions (e.g., ku sonyen ‘that (or the) boy’) in anaphoric contexts. 47As a piece of anecdotal evidence, I still remember the ﬁrst time I heard the word ku−at the age of 8 from a popular Korean pop song. I had to ask my mother what it meant. 48As noted in footnote 27, even when the pronoun ku is used in written contexts, it mostly takes a discourse referent rather than an intra-sentential one.

51 Kimyuna-ka chakyongha-n cwuelli-nun cenkuk J.Estina maycang-eyse Kimyuna-nom wear-adn jewelry-top nationwide J.Estina store-in manna-l swu iss-ta. meet-adn bn exist-decl ‘On the 11th, J.Estina has released a new picture of its brand muse, Yuna Kim. In this picture, Yuna Kim wore exotic emerald-coloured jewelry [...] The jewelry worn by Yuna Kim in the summer advertisement is available in J.Estina stores nationwide.’ (http://stylem.mt.co.kr/stylemView.php?no=2017051110551135769type=1) (46) Nwuna-nun pwusan-eyse tayena-ss-ta. Nwuna-ka older.sister-top Pusan-at born-past-decl older.sister-nom kohyang-i-n pwusan-ul ttena-n kes-un kotunghakkyo-lul hometown-cop-adn Pusan-acc leave-adn that-adn highschool-acc colepha-kose-i-ess-ta. Ku hwu nwuna-nun tasi-nun pwusan-ey graduate-after-cop-past-decl that later older.sister-top again-top Pusan-to ka-ci mosha-yess-ta. kulena kohyang-ul hyangha-n go-AuxEnd cannot-past-decl but hometown-acc face-adn nwuna-uy ayceng-un hansi-to sik-un cek-i older.sister-gen affection-top moment-even cool-adn time-nom eps-ess-ta. not-exist-past-decl ‘My sister was born in Pusan. It was after graduating high school that my sister left hometown Pusan. After that, my sister couldn’t go back to Pusan again. But my sister ’s affection towards the hometown never cooled off for a moment.’ (N. R. Han 2006: 42, ex.(49), originally from I. S. Lee et al. 1997) (47) Yeca hanmyeng-i mwutay oynccok-eyse tulewa-ss-ta. talun yeca woman one-nom stage left-from enter-past-decl different woman hanmyeng-i mwutay olunccok-eyse tulewassta. Ku yeca-nun kkoch one-nom stage right-from enter-past-decl that woman-top flower pakwuni-lul tul-ko iss-ess-ta. basket-acc carry-comp prog-past-decl ‘A woman entered from stage left. Another woman entered from stage right. That (or the) woman was carrying a basket of flowers.’ (Ionin et al. 2012: 76, ex.(10b))

Given this context, it seems reasonable to conclude that child learners of Korean may not receive suﬃcient or clear evidence from the input language data with regard to the grammar of ku.49 49As noted in Section 2.3, this does not necessarily mean that the child learners are not able to establish bound variable anaphora as they might already have acquired it easily with the long-distance anaphor caki ‘self’ or null pronouns.

52 Following the two-grammar hypothesis by Han, Lidz, and Musolino (2007), I propose that such a paucity in input concerning the pronoun ku may be the source of the inter- speaker variation regarding the availability of its bound variable construal. According to Han et al.’s proposal, when the primary linguistic data that child learners of a given language are exposed to is compatible with (at least) two competing grammars, along with the lack of relevant input data that would otherwise assist them to choose between the two, they may have to choose one grammar at random. As a consequence, some learners may acquire one grammar and others may acquire another, thus resulting in the presence of two groups of speakers in the given language community.50 I argue that the situation of ku is similar to the situation described in Han et al. (2007). The input data regarding ku to child learners of Korean is compatible with two competing grammars, the “pronominal grammar” and the “demonstrative grammar”. The relevant input to aid the child learners to choose one grammar over the other is scarce, and so they must choose one grammar at random. Accordingly, some may acquire the “pronominal grammar” for ku, in which case it can be bound and be (co)referential as well, and others may acquire the “demonstrative grammar” for ku, in which case it can only be referential. I propose that this is why and how the inter-speaker variation in the bound variable construal of ku arises and exists in contemporary Korean.51 To the extent that these arguments based on the two grammar hypothesis are valid, it is naturally predicted that Korean speakers should show consistent and persistent behaviours

50Han et al.’s proposal is based on the experimental evidence of the inter-speaker variation in negation and quantiﬁer scope judgment in Korean, which is argued to be compatible with two competing grammars: verb-raising grammar and non-verb-raising grammar. As Korean is a head-ﬁnal language, they argue, the input data provide little indication of (non-)verb raising to child learners, and the leaners thus have to choose one grammar at random. 51Costa and Martins (2011) report that there is inter-speaker variation regarding the availability of contrastive focus fronting (CFF) in contemporary European Portuguese (EP). Consider the following data (the expression in boldface is a fronted constituent).

(i)A: Estas a.dizer que ele te disse uma mentira? are-2sg saying that he you-dat told a lie ‘Are you saying he lied to you?’ B: Pois disse. E com isso me convenceu. indeed told-3sg and with that me convinced ‘Indeed, he did. And so he convinced me’ B0:* Pois disse. E com a mentria me convenceu. indeed told-3sg and with the lie me convinced For some speakers of EP (including the ﬁrst author of the study), B and B0 are both acceptable while for others (including the second author), only B is acceptable. Costa and Martins (p.218) attribute this split to the existence of two distinct grammars regarding CFF. That is, some EP speakers have a more restrictive grammar that only allows fronting of constituents with [deictic] feature (labelled as ‘Grammar B’), but others have a less restrictive grammar that involves no such restriction (labelled as ‘Grammar A’). It is, of course, not clear at this point whether the situation of CFF is similar to that of ku in terms of a deﬁciency in the relevant input data, compatible with the two grammar hypothesis. Their argument, however, can be taken to provide theoretical support to the current proposal in the sense that it shares the idea that inter-speaker variation regarding the availability of a given linguistic phenomenon may be derived from the co-existence of two competing grammars.

53 with regard to the bound variable construal for ku, because they would each maintain one single grammar over time, “pronominal grammar” or “demonstrative grammar” (cf. Han et al. 2016). This prediction has already been confirmed by the findings of Experiment 2, which showed that the participants exhibited the same judgments on the bindability of ku across test sessions separated by one month. Additionally, it can reasonably be postulated that there exist two kinds of pronoun ku: one which can be construed as a bound variable (“pronominal grammar”) and one which cannot be so construed (“demonstrative grammar”). If this is so, another important question then arises: how can the two kinds of ku with contrasting binding possibilities be accounted for in binding-theoretic terms? I attempt a proposal based on the work of Déchaine and Wiltschko (2002). Recall that Déchaine and Wiltschko propose three types of pronouns, which respectively have different maximal projections: Pro-DP, Pro-FP, and Pro-NP. This three-way distinction in pronouns is fundamentally distinguished from the structural approaches of Noguchi (1997) and Panagiotidis (2002), wherein two types of pronouns (D-pronoun and N-pronoun) are placed at different syntactic heads, D and N, but are uniformly of the same syntactic category DP. It has been discussed that such a fundamental contrast between the structural approaches is a reflection of their different underlying views on whether distinct binding-theoretic properties must be represented by distinct syntactic projections (Déchaine and Wiltschko) or distinct internal structures (Noguchi, Panagiotidis). I follow Déchaine and Wiltschko in assuming that the grammar of binding cannot “look into” a pronominal structure, and therefore internal structural differences of pronouns cannot result in their “external” differences (i.e., binding possibilities). I follow their argument that the Binding Conditions are only sensitive to the outer layer (i.e., the maximal projection) of a given pronoun (see, however, Koak 2008 for a different point of view). Adopting Déchaine and Wiltschko’s (2002) typology of pronouns, I argue that the two types of ku exhibit distinct binding properties because they are of different syntactic categories. Recall that Déchaine and Wiltschko’s three pronoun types are argued to have different binding-theoretic status, as summarized below.

(48) a. Pro-DPs are demonstrably deﬁnite and function like R-expressions, subject to Condition C (e.g., (Upriver) Halkomelem tú-tl’ò ‘he’ and thú-tl’ò ‘she’). b. Pro-FPs behave as variables, equal to the classical “Condition B pronouns” (e.g., English he and she, Shuswap newí7-s ‘he or she’). c. Pro-NPs are inherently constants, undeﬁned with respect to binding theory (e.g., Japanese kare ‘he’ and kanozyo ‘she’).

First, one type of ku is of the category FP, as illustrated in (49), with the binding-theoretic status of an ordinary “Condition B pronoun” such as English he. It can thus be used as a bound variable or a free variable.

54 (49) Pro-FP ku FP

F NP

ku ∅ The other ku is of the category DP, as illustrated in (50), and has the binding-theoretic status of an R-expression (see Kwon et al. 2009 for a similar argument that ku functions as an R-expression).52 The Pro-DP ku, subject to Condition C, cannot be construed as a bound variable, and can only refer.

(50) Pro-DP ku DP

D FP

ku F NP

∅ Under the assumption that the Korean demonstrative determiners, ku ‘that’, ce ‘that over there’, and i ‘this’, are heads of DP (M.Y. Kang 2001, Chang 2009, among many others), the Pro-DP ku shares the same syntactic structure with the demonstrative ku, and thus can be claimed to be a demonstrative pronoun (i.e., a demonstrative determiner used pronominally with a null NP).53

52Much work on Korean nominals assume that they have a DP projection, following M. Y. Kang (2001), Ahn and Cho (2006), Chang (2009), and many others. See, however, e.g., Im (1998), J. B. Kim (2016), for an argument that there are only NPs, not DPs, in Korean and the demonstratives are placed in Spec-NP, instead. 53Similar to Halkomelem, Korean has morphologically complex third-person pronouns, which are composed of demonstrative determiners followed by a variety of bound nouns denoting ‘person’, as illustrated in (i). Following Oh (2010), I will refer to these pronouns as ‘quasi-pronouns’ (note that Sohn 1999 uses the term ‘compound pronouns’, instead). (i) Quasi-pronouns in Korean (based on Oh 2010: 1222) Style Singular Plural Child or Adult-Plain D-ay D-ay-tul Adult-Familiar D-i D-i-tul Adult-Polite D-pwun D-pwun-tul Male-Derogatory D-nom D-nom-tul Female-Derogatory D-nyen D-nyen-tul (D=demonstratives i, ku, and ce)

55 This Pro-DP analysis of ku is cross-linguistically supported by German demonstrative pronouns, which are argued to be Pro-DPs in Déchaine and Wiltschko (2002) and Baltin et al. (2013). In (spoken) German, the deﬁnite articles der, die, and das can be used as demonstrative pronouns that are inﬂected for case, gender, and number (Wiltschko 1998; Bosch et al. 2003). Much like the proposed Pro-DP ku, the German demonstrative pronouns can refer to individuals, as in (51a) and (51b), but cannot be construed as bound variables, as in (52). Note that in contrast to the demonstrative pronouns, the corresponding personal pronouns in German (ihn, sie, er), which are analyzed as Pro-FPs in Déchaine and Wiltschko and Baltin et al., can be bound as well as referential.

(51) a. Maria hat {ihn/den} gesehen. Mary has {him/dem} seen ‘Mary has seen him.’ (Wiltschko 1998: 144, ex.(1))

Within the framework of Déchaine and Wiltschko’s (2002) pronominal typology, the Korean quasi-pronouns can be analyzed as pro-DPs, as illustrated below, where the demonstrative part is placed in D and the nominal part in N (cf. Koak 2008).

(ii) DP

D FP

i/ku/ce F NP

ay/pwun/...

This pro-DP analysis would then lead us to expect that the ku-based quasi-pronouns may receive a referential reading, but not a bound variable reading. This expectation is in fact borne out, as the following examples illustrate (kyay in (iiia) and (iva) is a contracted form of ku-ay).

(iii) a. Chinkwu1-ka kyay1 chinkwu cip sacin ccik-ese ponay-cwu-ess-e. friend-nom he friend house picture take-and send-give-past-decl

‘My friend1 took a picture of his1 house and sent it to me.’

b. Hanunim1-kkeyse kupwun1-i sinloyha-l swu iss-nun salam-tul-ul sayongha-si-ess-ta. God-hon.nom he-nom trust-adn can exist-adn person-pl-acc use-hon-past-decl

‘God1 used people who he1 could trust.’ (Google)

(iv) a. Nay motun chinkwu1-ka Mina-ka kyay∗1/2-lul coaha-yss-ta-ko sayngkakha-yss-e. My every friend-nom Mina-nom he-acc like-past-decl-comp think-past-decl

‘Every friend of mine1 thought Mina liked him∗1/2.’

b. Enu sensayng1-nim-ina kupwun∗1/2-ul conkyengha-nun haksayng-ul coaha-n-ta. every teacher-hon-also he-acc respect-adn student-acc like-pres-decl

‘Every teacher1 likes a student who respects him∗1/2.’ (adapted from Koak 2008)

As noted by researchers such as Cho 1992, H. Y Kim 1989, and Oh 2010, in contrast to the pronoun ku, the quasi-pronouns (especially, the Child/Adult-Plain style) are quite frequently used in both spoken and written Korean. Therefore, in the context of the current discussion, it can reasonably be said that with suﬃcient relevant linguistic input, child learners of Korean would all acquire the pro-DP quasi-pronouns, which would uniformly employ “demonstrative grammar”.

56 b. Paul sah eine Frau hereinkommen. {Sie/Die} trug einen Paul saw a woman enter {she/dem} wearing a schwarzen Mantel. black coat

‘Paul saw a woman1 enter. She1 was wearing a black coat.’ (Bosch et al. 2003: 1, ex.(2))

(52) Jeder Mann1 glaubt, daβ {er1/*der1} dumm ist. every man believes that {he/dem} stupid is

‘Every man1 believes that he1 is stupid.’ (Wiltschko 1998: 144, ex.(3))

Returning to the question of why and how the phenomenon of inter-speaker variation emerges and exists, we can now attempt to provide a more principled answer as follows. There exists two types of ku, Pro-DP ku and Pro-FP ku, in Korean.54 Given the lack 54The idea that a pronominal in a given language may be categorized into two structurally diﬀerent types is not new in the literature. As brieﬂy mentioned in footnote 11, in contrast to the standard observation that Japanese third-person pronouns can refer but cannot be bound, as shown in (i) and (ii), some researchers argue that the pronouns can have a bound variable reading in certain contexts, as shown in (iii).

(i) Taroo1-wa kare1-ga daihyoo-ni erabareru to omotte-iru. Taroo-top he-nom representative-as be.selected comp think-pres

‘Taro1 thinks that he1 will be selected as a representative.’

(ii)?* Dono gakusei1-mo kare1-ga daihyoo-ni erabareru to omotte-iru. every student-part he-nom representative-as be.selected comp think-pres

‘Every student1 thinks that he1 will be selected as a representative.’ (Yashima 2015: 1433, ex.(45a) and (45b))

(iii) Dono nooberusyoo zyusyoo sakka1-ga kare1-no kuruma-de kita-no? which Nobel.Prize winning author-nom he-gen car-in came-pres

‘Which Nobel Prize winning author1 came in his car?’ (Hoji 1991: 142, ex.(32c)) To account for the above observations, Yashima (2015: 1433) claims that the Japanese pronouns are in fact epithets (e.g., the bastard, the idiot, and the damn thing in English), but can be divided into two types, each with a distinct structure, as illustrated below.

(iv) [DP kare/kanozyo (=epithet phrase)]

(v) [DP pro [AppositiveP kare/kanozyo (=epithet phrase)]] First, one type of kare/kanozyo is a full DP epithet phrase, as in (iv), which can be bound or coreferential only when “Condition B and the so-called anti-logophoricity constraint are simultaneously satisfied”(p.1423), compatible with Dubinsky and Hamilton’s (1998) theory of epithets in general. Thus, (ii) is ungrammatical since kare is bound by the “agent of the attitude report”, i.e., a logophoric center, and thus the anti- logophoricity constraint is not satisfied. (Condition B is satisfied, though.) (iii), on the other hand, is grammatical since the antecedent of kare is not an attitude holder, thus satisfying the anti-logophoricity constraint as well as Condition B. The other type is an appositive epithet phrase justaposed with “a null pronominal anchor”, as in (v). Yashima argues that, being insensitive to anti-logophoricity constraint, this type of kare/kanozyo can be referential but not bound. Thus, in (i), kare can be coreferential with the matrix subject, a logophoric center (for further details of these arguments, see Yashima 2015: 1433-1434). A crucial point to be made here is that Yashima’s proposal theoretically supports the current proposal in the sense that it postulates two different syntactic structures for a given pronominal. Note, however, that his proposal implies that the two types of kare/kanozyo are acquired by all Japanese speakers, rather than each type being acquired by a distinct group of speakers.

57 of relevant input data, however, child learners of Korean must choose at random either of the two pronouns. That is, some may acquire the Pro-FP ku, which complies with the pronominal grammar and thus can be bound as well as be referential; others may acquire the Pro-DP ku, which complies with the demonstrative grammar and thus can only be referential.55 It follows from the current proposal that, in contrast to personal pronoun ku, the grammar of demonstrative ku should be uniform, as child learners of Korean are provided with suﬃcient input regarding demonstrative ku. An analysis consistent with the current proposal on personal pronoun ku given here would be to uniformly place demonstrative ku in the Pro-DP category of Déchaine and Wiltschko’s typology. This then predicts that all Korean speakers should disallow bound variable reading for deﬁnite descriptions with demonstrative ku. In light of observations found in the literature that English and other languages (Evans 1980; Hoji 1990; Nishigauchi 1990; Noguchi 1997; Elbourne 2008) allow a

55It is noteworthy at this point that just like Japanese pronouns kare and kanozyo, Korean pronouns ku and kunye can be preceded by adjectives, as illustrated in (i) (see again footnote 16 for the corresponding Japanese examples). In light of this observation, it might be tempting to argue that the Korean pronouns are nouns (e.g., Sohn 1994), and thus under Déchaine and Wiltschko’s (2002) pro-form typology, they should be treated as Pro-NPs (i.e., Ns contained within NPs) rather than Pro-DPs or Pro-FPs.

(i) a. twungtwungha-n ku b. yepkicek-in kunye fat-adn he bizarre-adn she

As pointed out to me by Chung-hye Han (personal communication, April 2017), however, the adjectives followed by the pronouns ku and kunye might better be analyzed as non-restrictive modiﬁers (see also Elbourne 2005: 163 for a similar argument for the Japanese cases). Indeed, (ia) and (ib) seem to mean ‘he, who is fat’ and ‘she, who is bizarre’ rather than ‘(lit.) fat he’ and ‘(lit.) bizarre she’, similarly to the example in (ii), where the adjective is understood as a non-restrictive relative clause modifying the nominal headed by the demonstrative ku. If this is the case, (i) does not constitute evidence that the Korean pronouns are lexical nouns, i.e., pro-NPs.

(ii) ketara-n [ku namwu] big-adn that tree ‘that (or the) tree, which is big’ (compare: ku ketara-n namwu ‘that big tree’) (Y. K. Kim 1997: 1)

Furthermore, the Korean pronouns are in fact not as noun-like as their Japanese counterparts are. First, ku and kunye cannot be modiﬁed by demonstratives, as in (iii), in contrast to kare and kanozyo.

(iii) a.*i ku b. * ce kunye this he that she

Note also that ku and kunye cannot mean ‘boyfriend’ and ‘girlfriend’, as in (iv), in contrast to kare and kanozyo, which do allow such a semantic shift.

(iv) a. Na-to hankukin {namcachinkwu/??*ku}-ka iss-ta. I-also Korean boyfriend-nom exist-decl ‘I also have a Korean boyfriend.’ b. Minsen-iga ni {yecachinkwu/??*kunye}-ni? Minsen-nom you.gen girlfriend-q ‘Minsen is your girlfriend?’

Based on the above discussion, I argue that the Korean pronouns should not be analyzed as pro-NPs.

58 bound variable reading for deﬁnite descriptions headed by a demonstrative in certain contexts (e.g., that logician), it remains an important task to test for the availability of such a reading in Korean. I return to this issue in Chapter 5, which is devoted to a discussion of future research.

59 Chapter 3

The syntax of Korean VP anaphora

3.1 Introduction

Cross-linguistically, it is possible to convey intended ideas without using full verbal expressions that would otherwise be necessary. A well-known example illustrating this fact is the English VP ellipsis construction given in (1). Here, a VP appears to have been omitted after will in the second conjunct (denoted by [VP ]), but is nevertheless readily interpreted as identical to the phonologically overt VP in the ﬁrst conjunct [VP happily eat those Paleo carrot ﬂatbread rounds with butter and jelly].

(1) The boys will [VP happily eat those Paleo carrot ﬂatbread rounds with butter

and jelly], and Boogie will [VP ], too. (www.unfancymama.com/20140301archive.html)

The Korean VP anaphora construction, which is the linguistic phenomenon of interest in the current chapter, is another case in point, as illustrated by the examples in (2) and (3).

(2) Wangca-tul-i [VP ku-eyke yelsimi kul-kwa yepep-ul paewu]-ess-ko, prince-pl-nom he-from hard writings-and manners-acc learn-past-conj Cwumong-to kuleha-yess-ta. Cwumong-also so.do-past-decl ‘The princes learned writings and manners hard from him, and Cwumong did so, too.’ (www.dev.emcampus.com)

(3)A:[ VP Eceypam-ey ku kongphoyenghwa-lul poko-naseo cam-ul last.night-at the scary.movie-acc watch-after sleep-acc selchi]-ess-e. bad.sleep-past-decl ‘I had a bad sleep last night after watching the scary movie.’ B: Na-to kulay-ss-e. I-also so.do-past-decl

60 ‘I did so, too.’

The VP anaphor kuleha ‘do so’ and its phonological variant kulay in (2) and (3) are immediately interpreted as identical to the longer and structurally more complex VPs in the

ﬁrst conjunct, [VP ku-eyke yelsimi kul-kwa yepep-ul paewu] ‘learn writings and manners hard from him’ and [VP eceypam-ey ku kongphoyenghwa-lul poko-naseo cam-ul selchi] ‘have a bad sleep last night after watching the scary movie’, respectively.1 The primary aim of this chapter is to provide novel empirical evidence that sheds new light on the syntactic status of the Korean VP anaphors, which is the key to understanding how they come to receive intended interpretations, despite the absence of otherwise expected lexical elements. As already mentioned in Chapter 1, the existing proposals on the syntax of Korean VP anaphora can be partitioned into two distinct ways. First, the VP anaphors can conceivably be analyzed as base-generated verbal pro-forms (Bae & Kim 2012; M. K. Park 2013). Under this view, kuleha in (2) or kulay in (3) can be taken to make reference to the event or situation introduced into the discourse by the VP in the ﬁrst conjunct, analogous to how an overt pronoun such as him in (4) may refer to the entity denoted by an antecedent DP.

(4) Mary hated [DP the old man who moved in from the east coast], but Lisa liked him very much!

In contrast, an alternative, and relatively more prevalent, view takes the VP anaphors to be instances of the ellipsis phenomena (e.g., Cho 1996; Son 2006; Storoshenko 2008; Kim & Yoon 2009; Ha 2010; Madigan 2015; M. K. Park 2015). To illustrate, consider the example of English VP ellipsis (1), which is repeated below in (5) (hereafter, strikethrough texts indicate elided material).

(5) The boys will [VP happily eat those Paleo carrot ﬂatbread rounds with butter and

jelly], and Boogie will [VP happily eat those Paleo carrot ﬂatbread rounds with butter and jelly], too.

1The VP anaphors kuleha and kulay can be interchangeably used in (2) and (3), with no apparent diﬀerence in meaning, as illustrated in (i) and (ii).

(i) Wangca-tul-i ku-eyke yelsimi kul-kwa yepep-ul paewu-ess-ko, Cwumong-to prince-pl-nom he-from hard writings-and manners-acc learn-past-conj Cwumong-also kulay-ss-ta. so.do-past-decl ‘The princes learned writings and manners hard from him, and Cwumong did so, too.’ (ii)A: Eceypam-ey ku kongphoyenghwa-lul poko-naseo cam-ul selchi-ess-e last.night-at the scary.movie-acc watch-after sleep-acc bad.sleep-past-decl ‘I had a bad sleep last night after watching the scary movie.’ B: Na-to kuleha-yess-e. I-also so.do-past-decl ‘I did so, too.’

61 With regard to the nature of the “missing VP” in the second conjunct, as Chung et al. (2011: 2) put it, “the standard view in current research in the Principles and Parameters framework and in the Minimalist Program” is what is generally referred to as the ‘PF- deletion’ approach (see also, among many others, Sag 1976, Chomsky & Lasnik 1993, Heim & Kratzer 1998, Tancredi 1992, Lappin 1996, Wilder 1997, Johnson 2001, Merchant 2001, Kennedy 2003, Büring 2005).2 The gist of it is as follows. A regular VP is constructed in overt syntax (e.g., happily eat those Paleo carrot...with butter and jelly in (5)), but may be “deleted” at the level of Phonological Form (PF) under identity with an antecedent VP, i.e., an overt VP elsewhere in the discourse.3 At the level of Logical Form (LF), however, the inaudible VP constituent remains intact with a fully represented structure, and thus it is this LF structure that gets passed to the semantic component to be assigned a proper semantic interpretation. In a similar vein, the Korean VP anaphora phenomena could also be analyzed as an output of the PF-deletion process. That is, a fully-articulated VP generated in overt syntax may get “replaced” (and thus “deleted”) at PF by certain surface forms such as kuleha or kulay, complying with “a pronunciation rule that applies only when [the] VP is identical to another one” in the discourse context (Cecchetto and Percus 2006: 79). Under this ellipsis analysis, the derivation of the second conjunct in (2), repeated below in (6), can be understood as is illustrated (7) (with irrelevant details suppressed).4

2I assume the standard (inverted) Y-model of grammar (Chomsky 1995, among others), according to which linguistic objects constructed in overt syntax are sent to the PF (an interface with the sensory-motor system) and the LF (an interface with the conceptual-intentional system). 3For expository purposes, I follow Sag (1976), Heim and Kratzer (1998), Fox (2000), Kennedy (2003) in imposing a parallelism condition on (VP) ellipsis constructions requiring that the elided constituent has an antecedent constituent with an identical representation at LF. This syntactic identity requirement can be understood as stated in Heim and Kratzer (1998: 250).

i. LF Identity Condition on Ellipsis: A constituent may be deleted at PF only if it is a copy of another constituent at LF.

Alternative semantic accounts of parallelism/identity, such as Merchant’s (2001) E-GIVENness, are consistent with the proposals in this chapter. 4The VP anaphor kuleha consists of two components kule ‘so’ and ha ‘do’, but the issue of how they are syntactically distributed within a verbal projection does not fall within the scope of the current study. Throughout this chapter, kuleha is treated as if it is an indecomposable string that simply corresponds to a VP, as shown in (7b) and (9b). See, however, Park (2015) for the claim that ha, as a light verb, comes under v, and it takes a VP complement that is replaced by kule, as shown in (i). See also Stroik (2001) and Hallman (2004) for a similar analysis of do so in English, as shown in (ii).

(i) vP (ii) vP

0 0 DPsubj v DPsubj v

VP v v VP

kule ha do so

(Park 2015: 704, ex.(24)) (Hallman 2004: 108, ex.(26b))

62 (6) ..., Cwumong-to kuleha-yess-ta. ..., Cwumong-also so.do-past-decl ‘(The princes learned writings and manners hard from him, and) Cwumong did so, too.’

(7) VP Ellipsis

a. Overt syntax and LF:

0 DPsubj T

Cwumong-to VP T

PP V0

‘from him’ AdvP V0

‘hard’ DPobj V

‘writings and manners’‘learn’ b. PF: TP

0 DPsubj T

Cwumong-to VP T

kuleha

Note above that a richly articulated VP structure does exist at the syntactic levels of representation (i.e., overt syntax and LF), albeit obscured by the surface form kuleha at PF. Thus, it is from this syntactic structure in (7a) that the intended meaning is derived. In Section 3.2 and Section 3.3, however, I present empirical data that contradict the ellipsis account of Korean VP anaphora. In two experimental studies to diagnose the presence of ‘hidden’ syntactic structure within the VP anaphors, I examined whether they license sloppy identity readings (Experiment 3 in Section 3.2) and overt extraction (Experiment 4 in Section 3.3). I demonstrate that the experimental results from the two studies could only be obtained if no deletion of syntactic structure was involved in deriving the VP anaphors.

For simplicity, the current paper eschews the vP/VP distinction, and vP-internal subjects.

63 On the basis of these empirical ﬁndings, I defend the claim that the Korean VP anaphors begin their lives as pro-forms in the derivation, and thus can accommodate no more structure than what must be present for the surface lexical string, and retrieve their semantic values from the context through interpretive rules, as in the case of pronominal resolution. To the extent that this VP pro-form analysis is on the right track, the second conjunct of (2), repeated below in (8), should be represented as in (9).

(8) ..., Cwumong-to kuleha-yess-ta. ..., Cwumong-also so.do-past-decl ‘(The princes learned writings and manners hard from him, and) Cwumong did so, too.’

(9) VP pro-form

a. Overt syntax and LF: b. PF:

TP TP

0 0 DPsubj T DPsubj T

‘Cwumong’ VP T ‘Cwumong’ VP T

kuleha kuleha

3.2 Experiment 3: Sloppy identity reading in Korean VP anaphora

3.2.1 Theoretical background

Since at least the classic work of Sag (1976), it has become a widespread practice in the literature to take the availability of a sloppy reading as an indication of the presence of unpronounced syntactic structure (i.e., the presence of ellipsis). Consider the English VP ellipsis construction in (10), in which the antecedent VP in the ﬁrst conjunct contains a possessive pronoun.

(10) John will [VP drink his beverage], and Mike will [VP ] too.

Given that the ﬁrst conjunct is intended to mean “John will drink John’s beverage”, the sentence as a whole may give rise to two diﬀerent interpretations, as illustrated in (11), which have been referred to as the sloppy identity and the strict identity readings since Ross (1967: 348).

64 (11) a. ‘John will drink John’s beverage, and Mike will drink Mike’s beverage, too.’ [sloppy identity] b. ‘John will drink John’s beverage, and Mike will drink John’s beverage, too.’ [strict identity]

In the sloppy reading in (11a), both John and Mike will drink their own beverage, while in the strict reading in (11b), both John and Mike will drink the beverages that belongs to the same individual, John.5 In Sag (1976) and much subsequent work inﬂuenced by it (e.g., Reinhart 1983; Heim & Kratzer 1998; Büring 2005), the sloppy−strict ambiguity is ascribed to a pronoun embedded within the ellipsis site [VP ], behaving as either a bound pronoun or a free pronoun. Specif- ically, according to their proposals, the antecedent VP in the ﬁrst conjunct in (10) may be represented at the level of LF in two distinct ways, depending on whether the pronoun his is construed as a bound variable or a free variable, and the elided VP in the second conjunct has−and must have−an LF representation that is identical to that of the antecedent VP, as shown in (12) and (13).6

(12)TP TP

DP DP λ1TP λ1TP John Mike 0 0 t1 T t1 T

T VP T VP

will V DP will V DP

drink his1beverage drink his1beverage

5If the owner of the beverage that John will drink is Mike or an individual other than John and Mike, say Robert, then two other strict identity readings can also be available, as illustrated in (i) and (ii).

(i) ‘John will drink Mike’s beverage, and Mike will drink Mike’s beverage, too.’ (ii) ‘John will drink Robert’s beverage, and Mike will drink Robert’s beverage, too.’

6For further details of this line of explanation, see e.g., Heim and Kratzer 1998 and Büring 2005.

65 (13)TP TP

DP T0 DP T0

John T VP Mike T VP

will V DP will V DP

drink his1beverage drink his1beverage

(12) illustrates an LF that would license the sloppy reading in (11a). In the first conjunct, quantifier raising (henceforth, QR) has been applied to the DP subject John, introducing a variable binder (or lambda operator), λ1. The pronoun his1 in the antecedent VP is then co- indexed with and c-commanded by that variable binder, and is thus semantically bound, i.e., a bound variable (Heim & Kratzer 1998).7 In the semantic component, the sister constituent of the QR’d DP is represented as a lambda function, in which each occurrence bearing the index 1 (i.e., the trace and the pronoun) is construed as a variable whose value is bound to that of the argument passed to the function, in this case, the subject of the first conjunct; therefore, the identity of the overt pronoun his1 is the individual John. Likewise, a variable binding configuration is also established in the second conjunct, wherein the pronoun his1 in the elided VP would be semantically bound by a variable binder. A lambda function is computed here as well, where the subject of the second conjunct, not the first conjunct, is taken as an argument, and therefore the elided pronoun his1 denotes the individual Mike. On the other hand, the LF structure in (13) would license the strict reading in (11b). Unlike (12), no variable binder is introduced in either conjunct due to the non-application of QR; subsequently, his1 in the antecedent VP and his1 in the elided VP are both free

7Heim and Kratzer’s (1998) deﬁnition of ‘semantic binding’ can be stated as below (paraphrased by McKillen 2016: 4).

(i) β semantically binds α if and only if: a. α is an occurrence of a variable b. β is an occurrence of a variable binder c. β c-commands α d. β is co-indexed with α e. β does not c-command any other variable binder occurrence which also c- commands and is co-indexed with α

Heim and Kratzer (1998: 263) also provide a derivative notion of semantic binding, as stated in (ii). (ii)ADP α semantically binds a DP β (in the derivative sense) iﬀ β and the trace of α are (semantically) bound by the same variable binder.

Given this derivative notion, the subject of each conjunct in (12) can be said to semantically bind each occurrence of the pronoun his1, although “the real binder” is actually the variable binder λ1.

66 variables. In other words, they are both considered to be semantically free and their values are determined contextually. Therefore, given the context where the index 1 is mapped to the individual John, the overt and elided occurrences of his1 will both be interpreted as coreferential with John, the subject of the ﬁrst conjunct. Based on the foregoing discussion, it could be said that sloppy identity readings are licensed in (VP) ellipsis contexts because the ellipsis site is the result of the deletion of a fully represented syntactic structure at PF, and thus there exists a pronoun at the ellipsis site at LF to be construed as a bound variable, in parallel with the overt bound pronoun in the antecedent clause. However, if the ellipsis site contained no internal syntactic structure (i.e., it was some kind of silent pro-form (cf. Lobeck 1995)), then a pronoun would never be available within the ellipsis site to serve as a bound variable, and consequently sloppy readings would never arise. As mentioned previously, the availability of a sloppy reading has been widely taken as a ‘sign’ for an unpronounced syntactic structure hosting a bound variable element. For example, Takahashi (1994) analyzes the so-called sluicing construction in Japanese as being derived by TP deletion (preceded by wh-movement), citing the presence of its sloppy reading as a supporting argument. Consider (14) below.8

(14) John-wa zibun-ga naze sikarareta ka wakattei-nai-ga, Mike-wa [CP naze1 [IP John-top self-nom why be-scolded Q knows-not-but Mike-top why zibun-ga t1 sikarareta] -ka] wakatteiru. -Q knows ‘John doesn’t know why John himself was scolded, but Mike knows why.’ (Takahashi 1994: 268, ex.(11))

The second conjunct containing sluicing can mean either that Mike knows why Mike himself was scolded, the sloppy reading, or that Mike knows why John was scolded, the strict reading. Takahashi (1994) argues that the sloppy reading is available in (14) because the sluicing site consists of an internally structured, but PF-deleted, TP constituent, where the anaphor zibun ‘self’ is construed as a bound variable. It is noteworthy, however, that a number of researchers have cautioned against taking the availability of a sloppy reading as a decisive indicator of the presence of ellipsis by demonstrating that sloppy readings can even be licensed in non-elliptical sentences such as those in (15), where overt pro-forms are involved (e.g., Fiengo & May 1994; Hoji 1998; 2003; Depiante 2000; Culicover & Jackendoﬀ 2005; Dalrymple 2005; Houser 2010; Frazier 2013; Merchant 2013a, 2013b; Kasai 2014).9

8See Takahashi (1994) for his argument that sluiced remnant in Japanese undergoes wh-movement to SpecCP. 9Merchant (2013a: 540) also provides pro-form examples such as (i)-(iii) below, where, he argues, “ellipsis cannot be implicated”, but sloppy readings are still found. Based on these examples, Merchant claims that the presence of sloppy readings is a “non-diagnostics” or “problematic diagnostics” for ellipsis.

67 10 (15) a. Betty cleaned her living room, and Jane [VP did the same thing], too. (compare: Betty cleaned her living room, and Jane did, too.) (i) ‘..., and Jane cleaned her own living room.’ [sloppy identity] (ii) ‘..., and Jane cleaned Betty’s living room.’ [strict identity] (Frazier 2013: 499, ex.(12a))

b. Max hit his friend, and Oscar [VP did it], too. (compare: Max hit his friend, and Oscar did, too.) (i) ‘..., and Oscar hit his own friend.’ [sloppy identity] (ii) ‘..., and Oscar hit Max’s friend.’ [strict identity] (Houser 2010: 15, ex.(31), originally from Fiengo & May 1994)

Given that do the same thing and do it are arguably verbal pro-forms that are syntactically atomic, i.e., they have no additional structure other than the one represented by the overt lexical items (Hankamer & Sag 1976), it is diﬃcult to maintain that the sloppy readings in (15) are resulted from a bound variable pronoun, which would be non-existent inside do the same thing and do it. Rather, the sloppy readings should be attributed to other sources such as semantic or pragmatic inferences, as put forward by Hoji (1998, 2003).11 For instance, do the same thing in (15a) may induce a sloppy reading by referring to the event or situation, say cleaning one’s own living room, that is inferred from the antecedent VP. Note also that in (16) below, do the same thing and do it allow sloppy readings even when there are no explicit linguistic antecedents in the discourse, further supporting that in pro-form contexts, sloppy interpretation can be resolved without recourse to a bound variable element.

(16) a. (Observing John touch his ﬁnger to his nose)

Bill [VP did the same thing]. (i) ‘Bill touched his own ﬁnger to his own nose.’ [sloppy identy] (ii) ‘Bill touched John’s ﬁnger to John’s nose.’ [strict identy] (Hoji 2003: 176, ex.(14))

(i) Harvey stubbed his toe on the doorstop, and it happened to Max, too. (ii) Undergraduates can be covered under their parents’ health plans if desired; {likewise for graduate students. | that goes for grad students, too.} (iii) A professor who pays down her mortgage with her paycheck is wiser than one who gambles it away in online poker. (Merchant 2013a: 540, ex.(7b-d)) 10According to Frazier (2013), it was found in Clifton and Frazier’s (forthcoming) written interpretation study that sentences containing do the same thing such as (15a) are even more likely to induce sloppy readings than their ellipsis counterparts. In light of this ﬁnding, Frazier (2013: 499) concludes that “sloppy readings may have many and varied sources” other than the variable binding mechanism, and therefore “sloppy readings do not diagnose ellipsis”. 11 The sloppy readings that are derived by a mechanism other than variable binding are not genuine sloppy readings, but “sloppy-like readings”, in the sense of Hoji (1998).

68 b. (John touches his ﬁnger to his nose. John says to Bill:)

Now you [VP do it]! (i) ‘Bill touches his own ﬁnger to his own nose.’ [sloppy identity] (ii) ‘Bill touches John’s ﬁnger to John’s nose.’ [strict identity] (Dalrymple 2005: 36, ex.(17))

It therefore follows that, as emphasized by Depiante (2000: 34), the availability of a sloppy reading is not, and should not be, “a necessary or suﬃcient condition” for determining the presence of ellipsis in that pro-forms can also induce sloppy interpretation; if an anaphoric element exhibits a sloppy reading, it might either involve deleted material or be an instance of a pro-form.

3.2.2 Research question and predictions

The above discussion clearly indicates that it is problematic to take the availability of a sloppy reading as a diagnostic for ellipsis in general. However, I will demonstrate that, with the aid of the overt third-person pronoun ku ‘he’, the (un)availability of a sloppy reading can be employed as a reliable tool to identify the syntactic nature of Korean VP anaphora, i.e., whether it is an instance of VP ellipsis or VP pro-form. We have already seen in Chapter 2 that the pronoun ku has a distinctive interpretative status from pronouns of other languages like English. In Experiments 1A, 1B, and 2, it was revealed that there was considerable inter-speaker variation in the acceptance of the bound variable construal of ku. That is, with regard to the interpretation of quantiﬁcational sentences such as (17), one group of Korean speakers consistently accepted a bound variable reading for ku (17a), while another group of Korean speakers consistently rejected such a reading. This is in sharp contrast to the English his in similar sentences which uncontroversially allows a bound variable reading.

(17) Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’

a. ‘Each person drank his own beverage.’ b. ‘Everyone drank one particular person’s beverage.’

On the basis of these empirical ﬁndings, it was proposed that the inter-speaker variation is indeed a robust linguistic phenomenon in present-day Korean, resulting from the co- existence of two competing grammars of ku; some Korean speakers may acquire one grammar for ku where it can be construed as a bound variable, but others may acquire another for ku where it cannot be so construed. Recall also from Experiments 1A, 1B, and 2 that ku was readily interpreted as coreferential with a referring expression elsewhere in the sentence, as illustrated in (18a), or as referring to some individual in the extra-linguistic context, as illustrated in (17b) and (18b).

69 (18) Minswu-ka ku-uy umlyoswu-lul masi-ess-ta. Minswu-nom he-gen beverage-acc drink-past-decl ‘Minswu drank his beverage.’

a. ‘Minswu drank Minswu’s beverage.’ b. ‘Minswu drank one particular person’s beverage.’

Building upon ﬁndings regarding the nature of ku, Experiment 3 was designed to address the following research question.

(19) Research question: Does the distribution of sloppy reading for VP anaphora follow from the distribution of the bound-variable pronoun in Korean?

In order to tackle this research question, the experiment examined the availability of sloppy readings in VP anaphora sentences as in (20) and bound variable readings in quantiﬁcational sentences as in (17).

(20) Minswu-ka [VP ku-uy umlyoswu-lul masi]-ess-ko, Kiswu-to Minswu-nom he-gen beverage-acc drink-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl ‘Minswu drank his beverage, and Kiswu did so, too.’

a. ‘Minswu drank Minswu’s beverage and, Kiswu drank Kiswu’s beverage, too.’ [sloppy identy] b. ‘Minswu drank Minswu’s beverage and, Kiswu drank Minswu’s beverage, too.’ [strict identy]

Given the context where ku in the first conjunct of (20) is understood as Minswu, the ellipsis and pro-form analyses of Korean VP anaphora yield different predictions regarding the relation between the distribution of the sloppy readings for kuleha and the quantificational binding of ku. This is so since the ellipsis analysis assumes kuleha to have syntactically accessible internal structure while the pro-form analysis does not. The predictions of the two analyses are given in detail below.

Predictions of the VP ellipsis analysis

If the VP anaphor kuleha involves PF-deletion of a fully articulated VP structure that is identical to its antecedent, then the pronoun ku must be present inside the VP anaphor, as represented in (21), and its (un)bindability would thus remain intact, which has been shown in Experiments 1A, 1B, and 2 to reveal substantial variation across Korean speakers.

70 (21)TP TP

DPsubj DPsubj λ1TP λ1TP Minswu-ka Kiswu-to

0 0 t1 T t1 T

VP T VP kuleha T

DPobj V DPobj V

ku-uy mwul-ul masi ku-uy mwul-ul masi

Accordingly, we would expect to find inter-speaker variation in the acceptance of sloppy readings in VP anaphora sentences as in (20) as well as in the acceptance of bound variable readings in quantificational sentences as in (17). And, more crucially, we should be able to observe a strong correlation between the distribution of the two readings. This would mean that an individual speaker’s acceptance of the sloppy reading in VP anaphora sentences should be predictable from her acceptance of the bound variable reading in quantificational sentences (and vice versa). For instance, speakers who allow the bound variable reading in (17a) would be expected to accept the sloppy reading in (20a), while speakers who do not allow the bound variable reading would be expected to reject the sloppy reading. Note that the strict reading in (20b), on the other hand, should be uniformly available to Korean speakers, since ku inside the VP anaphor would freely serve as a coreferential pronoun, which has been argued to be the source of strict identity under ellipsis.

Predictions of the VP pro-form analysis

If the Korean VP anaphor kuleha originates as an overt pro-form, then it would not accommodate an internal syntactic structure, and thus ku ‘he’ would never be present, as illustrated in (22).

71 (22)TP TP

DP DPsubj subj λ1TP λ1TP Minswu-ka Kiswu-to 0 t1 T

0 t1 T VP T

VP T kule-ha

DPobj V

ku-uy mwul-ul masi

Given that kuleha would never internally host the pronoun ku, the availability of its sloppy reading would not be contingent upon the (un)bindability of ku, and thus should not correlate with the availability of the quantiﬁcational binding of ku (as opposed to the predictions of the VP ellipsis analysis given above). Rather, given that pro-forms (e.g., English do it and do the same thing) can license sloppy readings through semantic or pragmatic inferences, we should expect the sloppy readings for kuleha to be uniformly available to Korean speakers. And we should reasonably expect the same for the strict readings for kuleha. In the following sections, the design and methodology of Experiment 3 are sketched, followed by the presentation and discussion of the results, which conﬁrm the predictions of the VP pro-form analysis.

3.2.3 Methodology

3.2.3.1 Participants

Forty-four native Korean adult speakers living in Vancouver, Canada, who did not participate in any of the previous experiments participated in Experiment 3. To avoid probable L2 eﬀects on L1 grammatical intuitions, the participants were required to not have lived in any foreign countries (including Canada) for more than 6 months.12 They were between the ages of 19 and 25 (the mean age was 22), and were all university students in Korea, but came to Canada to work part-time or study English at ESL institutions temporarily.

3.2.3.2 Task

A truth-value judgment task was employed, as in Experiments 1A, 1B, and 2. The participants were presented with sentences describing a context on a computer screen, followed by

12None of the participants had ever lived in a foreign country other than Canada. The mean period of their stay in Canada was 3 months at the time of participation.

72 Figure 3.1: Screenshot of a test trial in Experiment 3 a target sentence. They were then instructed to judge whether the target sentence truthfully described the given context by clicking 1 for ‘True’ and 0 for ‘False’ (see Figure 3.1 above).

3.2.3.3 Design and Materials

Each target sentence was either a VP anaphora sentence or a quantiﬁcational sentence, both of which contained ku as a possessive pronoun. Each context was biased to either bound or free interpretation of ku in the target sentence. Thus, two within-subjects factors were crossed to create four conditions: sentence type (VP anaphora (VPA) vs. Quantiﬁca- tional) × context type (Bound vs. Free). Consider a sample set of test items given in (23)-(26), where the target sentences are in boldface.13

(23) VPA-Bound (sloppy reading) condition:

Minswu, Kiswu, Cinswu-ka uwntong hwu swui-ko iss-ess-ta. Minswu Kiswu Cinswu-nom exercise after rest-prog past-decl Minswu-ka Minswu-uy umlyoswu-lul masi-ess-ta. Kiswu-to Minswu-nom Minswu-gen beverage-acc drink-past-decl Kiswu-also Kiswu-uy umlyoswu-lul masi-ess-ta. Kiswu-gen beverage-acc drink-past-decl

‘Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank Minswu’s beverage. Kiswu also drank Kiswu’s beverage.’

13A complete set of all test items can be found in Appendix D.

73 Minswu-ka ku-uy umlyoswu-lul masi-ess-ko, Kiswu-to Minswu-nom he-gen beverage-acc drink-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl

‘Minswu drank his beverage, and Kiswu did so, too.’

(24) VPA-Free (strict reading) condition:

Minswu, Kiswu, Cinswu-ka uwntong hwu swui-ko iss-ess-ta. Minswu Kiswu Cinswu-nom exercise after rest-prog past-decl Minswu-ka Minswu-uy umlyoswu-lul masi-ess-ta. Kiswu-to Minswu-nom Minswu-gen beverage-acc drink-past-decl Kiswu-also Minswu-uy umlyoswu-lul masi-ess-ta. Minswu-gen beverage-acc drink-past-decl

‘Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank Minswu’s beverage. Kiswu also drank Minswu’s beverage.’

Minswu-ka ku-uy umlyoswu-lul masi-ess-ko, Kiswu-to Minswu-nom he-gen beverage-acc drink-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl

‘Minswu drank his beverage, and Kiswu did so, too.’

(25) Quantificational-Bound condition:

‘Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank Minswu’s beverage. Kiswu also drank Kiswu’s beverage. Cinswu also drank Cinswu’s beverage.’

Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl

‘Everyone drank his beverage.’

(26) Quantificational-Free condition:

74 Minswu, Kiswu, Cinswu-ka uwntong hwu, Thayswu-ka oki-lul Minswu Kiswu Cinswu-nom exercise after Thayswu-nom coming-acc kitalimye, swui-ko iss-ess-ta. Minswu-ka Thayswu-uy umlyoswu-lul waiting rest-prog past-decl Minswu-nom Thayswu-gen beverage-acc masi-ess-ta. Kiswu-to Thayswu-uy umlyoswu-lul masi-ess-ta. drink-past-decl Kiswu-also Thayswu-gen beverage-acc drink-past-decl Cinswu-to Thayswu-uy umlyoswu-lul masi-ess-ta. Cinswu-also Thayswu-gen beverage-acc drink-past-decl ‘Minswu, Kiswu, and Cinswu were taking a rest after exercise, waiting for Thayswu to come. Minswu drank Thayswu’s beverage. Kiswu also drank Thayswu’s beverage. Cinswu also drank Thayswu’s beverage.

Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’

In (23), the context is compatible with the sloppy interpretation for kuleha in the target sentence, which would be, according to the ellipsis analysis, attributed to ku serving as a bound variable, while in (24), the context is compatible with the strict interpretation for kuleha in the target sentence, which would be attributed to ku being a free variable. In (25), the context is compatible with the quantificational binding interpretation for ku in the target sentence while in (26), the context is compatible with the referential (or deictic) interpretation for ku in the target sentence. The conditions that were key to the experiment were the VPA-Bound (sloppy reading) and the Quantificational-Bound conditions since the research question asks whether or not the distribution of sloppy readings for Korean VP anaphora patterns with the distribution of the bound variable construal of ku. The VPA- Free (strict reading) and Quantificational-Free conditions were taken to be controls, where ‘high’ acceptance rates were expected to be obtained, as discussed in Section 3.2.2. Sixteen sets of test items were constructed on the basis of the pattern illustrated in (23)-(26). The resulting 64 test items (16 items for each of the four conditions) were then distributed to four presentation lists using a Latin Square design, such that each list contained four items from each condition and, each participant saw exactly one version of each item. The same 40 filler items were then added to each list. Half of the fillers contained a target sentence which had a ‘True’ target truth-value, as illustrated in (27), and the other half contained a target sentence which had a ‘False’ target truth-value, as illustrated in (28).

(27) VPA-True condition: Kwangho, Changho, Yengho-ka yayengha-ko iss-ess-ta. Kwangho-ka Kwangho Changho Yengho-nom camping-prog past-decl Kwangho-nom swupak-ul mek-ess-ta. Changho-to swupak-ul mek-ess-ta. watermelon-acc eat-past-decl Changho-also watermelon-acc eat-past-decl

75 ‘Kwangho, Changho, and Yengho were camping. Kwangho ate watermelon. Changho also ate watermelon.’

Kwangho-ka swupak-ul mek-ess-ko, Changho-to kuleha-yess-ta. Kwangho-nom watermelon-acc eat-past-conj Changho-also so.do-past-decl

‘Kwangho ate watermelon, and Changho did so, too.’

(28) VPA-False condition:

Cinho, Kyuho, Dayho-ka swulcip-eyse iyakiha-ko iss-ess-ta. Cinho-ka Cinho Kyuho Dayho-nom bar-at talking-prog past-decl Cinho-nom maykcwu-lul masi-ess-ta. Kyuho-nun Vodka-lul masi-ess-ta. beer-acc drink-past-decl Kyuho-top Vodka-acc drink-past-decl

‘Cinho, Kyuho, and Dayho were talking at the bar. Cinho drank beer. Kyuho drank Vodka.’

Cinho-ka beer-lul masi-ess-ko, Kyuho-to kuleha-yess-ta. Cinho-nom beer-acc eat-past-conj Kyuho-also so.do-past-decl

‘Cinho drank beer, and Kyuho did so, too.’

3.2.3.4 Procedure

Experiment 3 was administered using Psychopy (Peirce 2007, 2009). All the participants began the experiment with six practice trials such that they could familiarize themselves with the required task. Sixteen test trials (four trials per condition) and 40 ﬁller trials were then presented in a uniquely generated random order. They could read the context and target sentences at their own pace, but they were instructed not to spend too much time on the truth-value judgement. On average, participants took 15-20 minutes to complete the entire experiment. They were paid $10 each as compensation for participation.

3.2.4 Findings

Out of a total of 44 participants, four participants were excluded from the data analysis because their accuracy rates on the filler trials were lower than 80%. Consequently, only results from 40 participants were analyzed in Experiment 3. Figure 3.2 below summarizes mean rates of acceptance (i.e., assignment of 1 ‘True’) by condition: 41% in the Quantificational-Bound condition, 80% in the VPA-Bound (sloppy reading) condition, 81% in the Quantificational-Free condition, and 79% in the VPA-Free (strict reading) condition. A generalized linear mixed-effects analysis of the participants’ responses was carried out using the lme4 function in the R statistical software; the factors sentence type and context type were entered as fixed effects, with participant and item as random effects. First, the analysis revealed a main effect of sentence type (estimated coefficient = 1.88, se

76 Quantificational VP Anaphora 100 80 60 40 Acceptance rate(%) Acceptance 20 0 Bound Free

Figure 3.2: Mean rates of acceptance and standard errors in Experiment 3

= .27, z = 6.91, p < .001). That is, regardless of context type, VP anaphora sentences are more likely to be accepted than quantificational sentences. Second, an interaction between sentence type and context type was found (estimated coefficient = −1.96, se = .40, z = −4.95, p < .001). That is, while speakers are equally likely to accept both bound and free readings for the VP anaphora sentences, they are more likely to accept free readings than bound readings for the quantificational sentences. Additionally, pairwise comparisons of mean acceptance rates in each of the conditions were conducted using a Tukey’s test. The results revealed that the acceptance rate in the VPA-Bound (sloppy reading) condition is significantly higher than the rate in the Quantificational-Bound condition (p < .001), while the acceptance rates in the VPA-Free (strict reading) and Quantificational-Free conditions are not different from each other. It was also revealed that the acceptance rate in the Quantificational-Free condition is significantly higher than the rate in the Quantificational- Bound condition (p <. 001), while the ratings in the VPA-Bound (sloppy reading) and the VPA-Free (strict reading) conditions are not different from each other. As in Experiments 1A, 1B, and 2, in order to investigate how the acceptance rate in the Quantificational-Bound condition was derived, all the participants were put into three different groups on the basis of their individual acceptance rates in that condition: accept (> 75% acceptance: assignment of 1 to three or four target sentences), Ambivalent (= 50% acceptance: assignment of 1 to two target sentences), and reject (< 25% acceptance: assignment of 1 to none or one target sentence). Consequently, a clear bimodal distribution

77 40 QuantBound VPABound 30 20 Number of participants Number 10 0 Reject(0-25%) Ambivalent(50%) Accept (75-100%)

Figure 3.3: Distribution of responses in Quantificational-Bound and VPA-Bound (sloppy reading) conditions in Experiment 3 was found, as illustrated in Figure 3.3 above. That is, 16 participants were assigned to accept while 23 participants were assigned to reject; there was only one participant who belonged to ambivalent. Given this assortment, the 41% of acceptance rate in the Quantificational-Bound condition does not actually mean that each participant accepted the quantificational binding interpretation for ku 41% of the time; rather, it can be seen as an indication that the 16 participants who are in the accept group, who also happen to make up 40% of the total number of participants, (almost always) consistently accepted the quantificational binding of ku. These findings thus replicate the inter-speaker variation reported in Experiments 1A, 1B, and 2: some Korean speakers allow ku to be construed as a bound variable while other Korean speakers do not.14 Additionally, the participants’ responses in the VPA-Bound (sloppy reading) condition were inspected in the same way as they were in the Quantificational-Bound condition.15 Crucially, the results revealed no such bimodal distribution as in the Quantificational-Bound condition, and thus no inter-speaker variation (see again Figure 3.3). That is, except for 4

14In a separate experiment of a similar design, which consisted of two sessions separated by one month, a clear bimodal distribution was found in the Quantificational-Bound condition in each session, with about 40% of the participants accepting the quantificational binding of ku. Moreover, a linear regression analysis revealed a strong correlation (r = 0.74, t = 6.43, p < .001) between the two sessions in the Quantificational- Bound condition. These results show that the participants exhibited the same judgment over time on the bound variable construal of ku, further supporting that the inter-speaker variation reported in the present study is indeed a synchronically active phenomenon in Korean. 15I am indebted to Jeff Runner for his suggestion of conducting this additional inspection.

78 ambivalent participants, the rest of the participants assigned 1 ‘True’ to three or four (out of four) target sentences (i.e., >75% acceptance), indicating that (nearly) all participants accepted the sloppy readings for the VP anaphors. This suggests that Korean speakers may be divided into two groups; those speakers who allow both bound variable readings for ku (in quantificational contexts) and sloppy readings for Korean VP anaphors, and those speakers who disallow the former, but allow the latter. This in turn suggests that the interaction between sentence type and context type is a result of the distinct acceptance behaviours in the Bound conditions between the two groups of participants. A linear regression analysis testing the correlation between the participants’ acceptance rates in the two Bound conditions revealed that the correlation coefficient is not significantly different from zero, indicating that an individual speaker’s acceptance of the sloppy readings for the VP anaphors is not predictable from her acceptance of the quantificational binding of ku (and vice versa), as illustrated in Figure 3.4 below. This result was expected since about half of the participants (represented by the dots in the dotted box in Figure 3.4) rejected the quantificational binding of ku but at the same time accepted the sloppy reading for the VP anaphors.

Figure 3.4: Correlation between mean acceptance rates in Quantiﬁcational-Bound and VPA- Bound (sloppy reading) conditions in Experiment 3

79 3.2.5 Discussion of Experiment 3

The findings of Experiment 3 provide empirical evidence that is incompatible with the view that Korean VP anaphors involve ellipsis. As discussed earlier, if kuleha in sentences such as (20), repeated in (29) below, was indeed derived by PF-deletion of a fully specified verb phrase housing ku, then the availability of its sloppy reading would be expected to correlate with the bindability of ku, which has been repeatedly found to exhibit inter-speaker variation through the examination of the bound variable readings in quantificational sentences such as in (17), repeated in (30) below.

(29) Minswu-ka [VP ku-uy umlyoswu-lul masi]-ess-ko, Kiswu-to Minswu-nom he-gen beverage-acc drink-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl ‘Minswu drank his beverage, and Kiswu did so, too.’ a. ‘Minswu drank Minswu’s beverage and, Kiswu drank Kiswu’s beverage, too.’ [sloppy identy] b. ‘Minswu drank Minswu’s beverage and, Kiswu drank Minswu’s beverage, too.’ [strict identy]

(30) Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’ a. ‘Each person drank his own beverage.’ b. ‘Everyone drank one particular person’s beverage.’

Contrary to the above expectation, however, Experiment 3 has found no correlation between the distribution of sloppy readings and that of bound variables. Specifically, it was found that in addition to a group of Korean speakers who accept both quantificational binding of ku (as in (30a)) and sloppy readings for kuleha (as in (29a)), there exists another group of Korean speakers who reject the former but, nevertheless, accept the latter; in other words, Korean speakers uniformly accept sloppy readings with Korean VP anaphors, irrespective of their acceptance or rejection of ku as a bound variable pronoun. As foreshadowed in the previous section, these findings are evidence that the VP anaphors are base-generated pro- forms. That is, they are atomic anaphoric elements that do not contain syntactic structure other than the lexical material that surfaces, and their interpretations are resolved by “a meaning recovery strategy similar to pronominal anaphora resolution” (Tomioka 2008: 210), which allows for semantic/pragmatic inferences to derive “sloppy-like” interpretations. Therefore, as long as relevant contexts are clearly provided, the sloppy readings for kuleha should be readily available, as confirmed in Experiment 3.16 16One might attempt to defend the ellipsis analysis of kuleha (thus, the presence of an internal structure), by postulating that those speakers who only allow ku to be (co)referential could nevertheless accept the

80 Before closing this section, note that the acceptance rates in the Quantiﬁcational-Free condition and in the VPA-Free (strict reading) condition are uniformly high. As discussed earlier, the high rate in the former was expected since it is commonly agreed that ku can readily induce referential (or deictic) readings as in (30b). The high rate in the latter was also expected since both the ellipsis and pro-form analyses predicted that strict readings as in (29b) should be readily available with the Korean VP anaphors. Therefore, these ﬁndings strongly suggest that the experimental design and test items were appropriate in assessing participants’ knowledge of anaphora, which, in turn, ensured the reliability of the results obtained in the two Bound conditions. The next section starts with a discussion of another diagnostic test that the present study has adopted to explore the syntactic status of Korean VP anaphora: the availability of overt extraction.

3.3 Experiment 4: Overt extraction out of Korean VP anaphora

3.3.1 Theoretical background

In the generative syntax literature, the availability of overt extraction has been widely taken to be a reliable tool to diagnose whether an (overt or null) anaphoric element involves ellipsis or pro-form resolution (Haïk 1987; Tancredi 1992; Fiengo & May 1994; Johnson 2001; Schuyler 2001; Haddican 2007; Houser 2010; Bentzen et al. 2013; van Craenenbroeck & Merchant 2013; Merchant 2013a, 2013b; Kasai 2014, among others). The principle that underlies this diagnostic test is quite straightforward: if overt extraction is allowed out of the site of a target anaphoric item, it would mean that the target item must involve a deletion of syntactic structure that can host the trace (or copy, in the Copy Theory of Movement) of the extracted phrase; by contrast, if overt extraction is not available to a target anaphoric item, it would mean that the target item must be a pro-form that is syntactically atomic, and thus is unable to host any traces of movements. In light of this, VP ellipsis in English has been shown to contain internal syntactic structure, on the basis sloppy interpretation for kuleha via the operation of ‘vehicle change’ (Fiengo & May 1994; Saﬁr 1999), “by which one type of expression (say, a name) can be [turned into] a distinct type of expression (say, a pronoun)” (S. W. Kim 1999: 270). Under this view, then, ku in the VPA site would transform to an LF object (α) that would have to be, for a sloppy reading, coreferential with the subject of the VPA clause, as illustrated in (i).

(i)[ TP Minswu1-ka [VP [DP ku1-uy umlyoswu-lul] masi]-ess]-ko, [TP Kiswu2-to [VP [DP α2-uy umlyoswu-lul] masi]-ess]-ta. ‘Minswu drank Minswu’s beverage, and Kiswu drank Kiswu’s beverage, too.’

Crucially, however, this idea is incompatible with Fiengo and May’s (1994: 218) claim that a nominal can only be ‘vehicle-changed’ into another form when its index value remains the same. Even if the alleged vehicle change could somehow derive an LF structure as in (i), PF-deletion (thus, ellipsis) would not be licensed since the indices assigned to the overt ku and the vehicle-changed entity are diﬀerent and therefore does not satisfy the LF identity requirement that the elided material be identical to its antecedent (see footnote 3).

81 of sentences as in (31a)-(31d), where the ﬁrst conjuncts all involve movement operations within the framework of (Chomskyan) generative grammar.17

(31) a. Indirect wh-question

I don’t know [DP which puppy]1 you should [VP adopt t1], but I know [DP which 18 one] you shouldn’t [VP ]. (adapted from Schuyler 2001: 1, ex.(1)) b. Direct wh-question

[DP Which ﬁlms]1 did you refuse to [VP see t1], and [DP which ﬁlms] did he

agree to [VP ]? (Merchant 2013b: 19, ex.(92a)) c. Relativization

This is the book [DP which]1 O.J. Berman [VP reviewed t1], and this is the one

[[DP which] Fred won’t [VP ]. (adapted from Johnson 2001: 18, ex.(62b)) d. Topicalization

[DP Potatoes]1 I[VP like t1], but [DP tomatoes] I don’t [VP ]. (van Craenenbroeck & Merchant 2013: 705, ex.(8b))

Let us focus on the sentence in (31a). Here, which one in the second conjunct, being readily interpreted as part of the indirect question that the verb know takes, can reasonably be viewed as a moved wh-phrase in an analogous way as which puppy in the ﬁrst conjunct. The origin of this moved wh-phrase is then easily accounted for if we postulate that there is an elided full-ﬂedged VP after shouldn’t embedding a trace of the moved wh-phrase. This is illustrated in (32).

17The possibility that the wh-phrase is base-generated in situ is not considered in the current study. See, however, Gazdar (1982), Pollard and Sag (1994), and Kim and Sells (2008) for a non-movement analysis of English wh-questions that posits feature percolation mechanism. 18Schuyler (2001: 18) argues that for extraction out of the VP ellipsis site to be allowed, “there must be a contrastively focused expression in the c-command domain of the extracted phrase”. In his original example in (31a), for instance, focal stress is placed on the auxiliaries should and shouldn’t, which is indicated by capitalization (e.g., SHOULD and SHOULDN’T). He judges sentences like that in (i) below to be ungrammatical, where none of focused elements are included in the c-command domain of the extracted phrases.

(i) *PETE knows [DP which puppy]1 you should [VP adopt t1], but JAN doesn’t know [DP which one] you should [VP ]. (Schuyler 2001: 10, ex.(59))

However, the alleged prosodic eﬀects is not considered in examining the availability of extraction out of Korean VP anaphora, and I will leave this issue for future investigation.

82 (32) CP CP

0 0 which puppy1 C which one1 C

C TP C TP

DP T0 DP T0

you T VP you T VP

should V t1 shouldn’t V t1

adopt adopt

In the structure for the second conjunct, which one has been extracted to the specifier of the CP from within the VP, the remnant of which ([VP adopt t1]) is subsequently deleted under identity with the antecedent VP in the first conjunct. If the ellipsis site were a null pro-form and thus could not contain any structure to host traces of any movements, the wh-phrase would end up violating Chomsky’s (1986) restriction on vacuous quantification (e.g., Johnson 2001; Schuyler 2001).19

19Crucially, whether an anaphor has a null phonological realization is irrelevant to whether it can license overt extraction or not. For instance, Null Complement Anaphora (NCA), which has been argued to be a null pro-form, does not pass the extraction diagnostics, as illustrated in (i)-(ii) below (e.g., Hankamer & Sag 1976; Depiante 2000; Merchant 2013).

(i) *I know [which book]1 Mary volunteered to read t1, and Peter knows which article Sally volunteered [NCA ]. (compare: I know Mary volunteered to read Harry Potter, and Peter knows Sally volunteered [NCA ], too.) (J. S. Kim 2010: 311, fn.14, originally from Depiante 2000)

(ii) *Which films did you refuse to see, and which films did he refuse [NCA ]? (compare: You refused to see those films, and he refused [NCA ], too.) (Merchant 2013b: 19, ex.(92b)) Japanese null clausal complement has also been claimed to be a null pro-form, resisting overt extraction out of it, as illustrated in (iii) below (e.g., Shinohara 2006; Tanaka 2008; Kasai 2014).

(iii) *Sono hon-o1 Taroo-wa [CP Hanako-ga t1 katta to] itta si, sono hon-o2 that book-acc Taroo-top Hanako-nom bought comp said and that book-acc Ziroo-mo [CP Hanako-ga t2 katta to] itta. Ziroo-also Hanako-nom bought comp said Indented: ‘Taroo said that Hanako bought that book, and Ziroo also said that she bought that book.’ (Kasai 2014: 181, ex.(37), originally from Saito 2007)

Note that the non-extraction counterpart in (iv) is totally acceptable.

(iv) Taroo-wa [CP Hanako-ga Sono hon-o katta to] itta si, Ziroo-mo [CP Hanako-ga Taroo-top Hanako-nom that book-acc bought comp said and Ziroo-also Hanako-nom sono hon-o katta to] itta. that book-acc bought comp said Indented: ‘Taroo said that Hanako bought that book, and Ziroo also said that she bought that book.’

83 Consider now the following examples, in which the elided VPs in (31a)-(31d) have been replaced with the VP pro-form do it.

(33) a.* I don’t know [ DP which puppy]1 you should [VP adopt t1], but I know which

one you shouldn’t [VP do it].

b.* [DP Which ﬁlms]1 did you refuse to [VP see t1], and which ﬁlms did he agree

to [VP do it]?

c.* This is the book [ DP which]1 O.J. Berman [VP reviewed t1], and this is the one

which Fred won’t [VP do it].

d.* [DP Potatoes]1 I[VP like t1], but tomatoes I don’t [VP do it].

The sentences above are all regarded as ungrammatical in contrast to their VP ellipsis counterparts, indicating that do it does not tolerate the intended movement operations. These extraction impossibilities can be easily accounted for if we assume that do it has no internal structure other than the one represented by the lexical items ‘do’ and ‘it’, and thus extractable elements are never internally existent (e.g., Johnson 2001; Schuyler 2001; Haddican 2007; Houser 2010), as illustrated in the following structure (34) for (33a).

(34) CP *CP

0 0 which puppy1 C which one C

C TP C TP

DP T0 DP T0

you T VP you T VP

should V t1 shouldn’t do it

adopt

Here, if the site of do it contained the syntactic material [VP [V adopt][DP which one]] in the early stages of derivation, we would then expect that the VP-internal object, which one, would undergo wh-movement to its surface position, and the remnant VP might be “masked” by do it at PF under identity with the antecedent VP in the ﬁrst conjunct. Under this reasoning, then, the whole second conjunct should be a legitimate output, which is contrary to fact. Turning now to Korean VP anaphora, note that Korean is generally assumed to have no overt wh-movement (i.e., a wh-in-situ language), but it exhibits “scrambling”, which is standardly analyzed as an overt movement process that derives non-canonical (i.e., scrambled) word orders (e.g., Ross 1967, Harada 1977, Saito 1985, 1992, Mahajan 1990, Fukui 1993,

84 Bailyn 2001; for Korean, see Y. S. Lee and Santorini 1994, Ko 2007, and E. S. Lee 2007).20 Consider (35) and (36) below, where the examples in (a) are sentences with the canonical SOV word order of Korean while the examples in (b) are their scrambled counterparts.21

(35) a. Yuri-ka [VP sakwa-lul mek]-ess-e. Yuri-nom apple-acc eat-past-decl ‘Yuri ate apples.’

b. Sakwa-lul1 Yuri-ka [VP t1 mek]-ess-e. apple-acc Yuri-nom eat-past-decl ‘Yuri ate apples.’

(36) a. Minswu-ka [VP [CP Yuri-ka sakwa-lul mek-ess-tako] Minswu-nom Yuri-nom apple-acc eat-past-comp sayngkakha]-yss-e. think-past-decl ‘Minswu thought that Yuri ate apples.’

b. Sakwa-lul1 Minswu-ka [VP [CP Yuri-ka [VP t1 mek]-ess-tako] apple-acc Minswu-nom Yuri-nom eat-past-comp sayngkakha]-yss-e. think-past-decl ‘Minswu thought that Yuri ate apples.’

According to the standard approach, the scrambled sentence in (35b) is derived from the canonical simple sentence in (35a) by the movement of sakwa-lul ‘apple-acc’ out of the VP across the subject; the scrambled sentence in (36b) is derived from the canonical complex sentence containing an embedded clause in (36a) by the movement of sakwa-lul ‘apple-acc’ out of the embedded VP across the matrix subject as well as the embedded subject. In the literature, the former type of scrambling is often called “short distance scrambling” (or “clause-internal scrambling”), since an element moves within the boundary of the clause it originates from, while the latter type is called “long distance scrambling” (or “clausal scrambling”) since an element moves across the clause boundary. Accordingly, to the extent that it is an overt movement operation, the phenomenon of scrambling could reasonably be employed to diagnose the syntactic status of the VP anaphora in Korean, in an analogous fashion as in the extraction tests done for English VP ellipsis and do it.22 That is, if the Korean VP anaphors allow scrambling, it would mean that they contain elided material; if they resist scrambling, however, it would mean that they are pro-forms with no internal syntactic structure. However, determining the (un)availability

20For base-generation (i.e., non-movement) accounts of scrambling, see Neeleman (1994) and Boskovic and Takahashi (1998). 21As can be known from their translations, the scrambled sentences and the canonical sentences share the same meaning (or proposition), though they might be used in diﬀerent pragmatic contexts since some kind of emphasis would be put on the scrambled object sakwa-lul ‘apple-acc’. 22Hereafter, the terms “extraction” and “scrambling” will be used interchangeably.

85 of scrambling out of the VP anaphors might not be as simple and straightforward as it appears to be. First, to the best of my knowledge, there have been no studies other than Park (2015) in the literature of Korean syntax addressing the issue in question with a set of relevant data. Moreover, according to Park’s (and presumably his informants’) introspective judgments, scrambling may or may not be possible out of the VP anaphors, depending on whether it is short distance or long distance scrambling, as illustrated in (37) and (38).

(37) a. A: Say-lul1 Chelswu-ka [VP t1 koylophi]-nta. bird-acc Chelswu-nom nag-decl. ‘Chelswu is nagging a bird.’ B: Cwi-lul Yenghuy-ka kulay. mouse-acc Yenghuy-nom so.do Intended: ‘Yenghuy is nagging a mouse.’ (adapted from Park 2015: 695, ex.(4))

b. Sakwa-lul1 nay-ka [VP mayil t1 mek]-ess-ko, photo-lul nay Apple-acc I-nom everyday eat-past-and grape-acc my tongsaying-i kulay-ss-e. brother-nom so.do-past-decl Intended: ‘I ate apples everyday, and my brother ate grapes everyday.’ (adapted from Park 2015: 698, ex.(11))

(38) A: LGB-lul1 Cheli-nun [VP [CP Yenghi-ka [VP t1 ilk]-ess-tako] LGB-acc Cheli-top Yenghi-nom read-past-decl sayngkakha]-yss-e. think-past-decl. ‘Cheli thought that Yenghi read LGB.’ B: *Barriers-lul Toli-nun kulay-ss-e. Barriers-acc Toli-top so.do-past-decl Intended: ‘Toli thought that Yenghi read Barriers.’ (adapted from Park 2015: 709, ex.(29))

Park (2015) reports that the second conjuncts in (37a) and (37b) are grammatical, where the objects in the sentence-initial position, cwi-lul ‘mouse-acc’ and photo-lul ‘grape-acc’, are taken to have been short-scrambled out of kulay. On the other hand, the second conjunct in (38) is ungrammatical, where the sentence initial object, Barriers-lul ‘Barriers-acc’, is taken to have been long-scrambled out of kulay. Assuming that the short distance scrambling is an A-movement operation while the long distance scrambling is an A-movement¯ operation (cf. Mahajan 1990; Saito 1992), Park claims that the VP replacement by kulay can only target a VP constituent containing an A-trace (or no trace at all),23 but not a VP constituent containing an A-trace,¯ thereby resulting in the asymmetry observed above.

23Park (2015) follows Lasnik’s (1999) proposal that A-movement, unlike A-movement,¯ does not necessarily leave an (A-)trace.

86 In brief, according to Park (2015), Korean VP anaphora is derived by ellipsis, and thus can permit overt extraction, but only when it is an instance of A-movement, e.g., short distance scrambling.24 Obviously, this is not compatible with the VP ellipsis analysis we have discussed so far, which predicts that both types of scrambling, in principle, could take place out of the VP anaphora. This also does not align with what follows from the VP pro-form analysis conﬁrmed by Experiment 3: if the VP anaphora is a pro-form, neither type of scrambling should be acceptable since any extractable syntactic objects would be non-existent to begin with.

3.3.2 Research question and predictions

Contrary to what has been observed by Park (2015), however, all of the eight Korean native speakers I informally consulted shared the same intuition that the short distance examples in (37) and the long distance example in (38) are both completely unacceptable.25 Experiment 4 was motivated by the discrepancy between Park’s (2015) and my informants’ acceptability judgments (or intuitions) on the availability of extraction from Korean VP anaphora. It addresses the following research question.

(39) research question: Is overt extraction out of the Korean VP anaphora available in both the short distance and long distance scrambling contexts? (Is the short−long scrambling asymmetry reported by Park (2015) a real linguistic phenomenon that exists in Korean?)

The experiment thus examined the acceptability of VP anaphora sentences involving short scrambling, as in (40), and those involving long scrambling, as in (41), with respect to which the ellipsis and pro-form analyses make diﬀerent predictions.

(40) A: Manhwa-lul1 na-nun [VP cacwu t1 ilk]-ess-e. Ne-nun? comic.book-acc I-top often read-past-decl you-top ‘I often read comic books. How about you?’

B: Manhwa-lul na-to kulay-ss-e. comic.book-acc I-also so.do-past-decl Intended: ‘I often read comic books, too.’

(41) A: Manhwa-lul1 na-nun [VP [CP Yuli-ka cacwu t1 ilk-ess-tako] comic.book-acc I-top Yuli-nom often read-past-comp tul]-yess-e. Ne-nun? hear-past-decl you-top

24In his earlier work, however, Park (2013: 56) argues that “kulay ‘do so’ anaphora” is a pro-form rather than an ellipsis phenomenon. 25We will see in the next section that these informant’s introspective judgments were replicated in Exper- iment 4.

87 ‘I heard that Yuli often read comic books. How about you?’

B: Manhwa-lul na-to kulay-ss-e. comic.book-acc I-also so.do-past-decl

Intended: ‘I also heard that Yuli often read comic books.’

Predictions of the VP ellipsis analysis

If the VP anaphor kulay is an instance of ellipsis, overt object scrambling should be possible in both the short distance and long distance contexts since there exists a syntactic structure that hosts an extractable object, as illustrated in (42a)-(42b) below.26

(42) a. TP

Manhwa-lul1obj TP

0 na-tosubj T

VP kulay T

t1 V

ilk b. TP

Manhwa-lul1obj TP

0 na-tosubj T

VP kulay T

CP V

Yuri-ka...t1...tako tul

26Although there are proposals in the extant literature that the landing sites for short and long-scrambled objects are not uniform (for example, Mahajan 1990), in this paper, we simply assume that both short and long-scrambled objects adjoin to TP, in the spirit of Saito (1992), for the sake of convenient illustration. Whether the landing site of short and long-scrambled objects are distinct is not directly pertinent to the current discussion.

88 Given the structures above, the surface string ‘(scrambled)Object−Subject−kulay’ should be readily acceptable to Korean speakers. Alternatively, if Park’s (2015) version of ellipsis analysis is indeed on the right track, we should be able to observe the short−long scrambling asymmetry.

Predictions of the VP pro-form analysis

If the VP anaphor kulay is a base-generated pro-form, and thus does not have an internal structure to host an extractable object, then overt object scrambling should not be possible in either the short distance or long distance contexts, as illustrated in (43) for both short distance and long distance scrambling.

(43) *TP

Manhwa-lulobj TP

0 na-tosubj T

VP T

kulay

Given the structures above, the surface string ‘(scrambled)Object-Subject-kulay’ should be unacceptable to Korean speakers. In the following sections, we sketch the design and methodology of Experiment 4, and then present and discuss its results, which are consistent with the predictions of the VP pro-form analysis.

3.3.3 Methodology

3.3.3.1 Participants

Forty-eight adult native speakers of Korean living in Vancouver, Canada, took part in Experiment 4. They all met the eligibility criteria for participation, which are identical to those for Experiment 3. They were between the ages of 20 and 25 (the mean age was 23), and they came to Vancouver to work part-time or study English. Except for eight participants, the rest were university students in Korea.27

27As in Experiment 3, all participants had never lived in any foreign countries other than Canada. The mean period of their stay in Canada was 5 months at the time of participation.

89 Figure 3.5: Screenshot of a test trial in Experiment 4

3.3.3.2 Task

A 7-point Likert scale task (e.g., Weskott and Fanselow 2011, Sprouse and Almeida 2011, Sprouse et al. 2013) was employed in Experiment 4. The participants were presented with a context sentence on a computer screen, followed by a dialogue between two people. The dialogue consisted of two sentences, the ﬁrst one played the role of an “antecedent sentence” and the second one was the target sentence. The participants’ task was to rate the acceptability of the target sentences by clicking on a number from 1 (labelled as “Unnatural”) to 7 (labelled as “Natural”) (see Figure 3.5 above).

3.3.3.3 Design and Materials

Each target sentence contained the VP anaphor kulay, and from the perspective of the ellipsis analysis, either had or didn’t have object scrambling to the sentence initial position.28 Each antecedent sentence had either a “simple VP” or a “complex VP” containing an embedded clause. These VPs served as an antecedent to the VP anaphor in the target sentence. Two within-subjects factors were thus crossed to create four conditions: antecedent VP type (Simple vs. Complex) × extraction (Extraction vs. NoExtraction). Consider a sample set of test items given in (44)-(47), where the target sentences are in boldface.29

28Unlike in Experiment 3, the VP anaphor kulay, not kuleha, was used for the target sentences in Experi- ment 4, because it is generally deemed to sound more natural or authentic in colloquial contexts. 29A complete set of all test items are provided in Appendix E.

90 (44) Simple-Extraction (short distance scrambling) condition:

Minswu-wa Hanswu-ka kakca etten kwamok-ul yelsimhi Minswu-and Hanswu-nom each which subject-acc hard kongpwuha-yess-nunci malha-ko iss-ess-ta. study-past-comp talk-prog past-decl ‘Minswu and Hanswu were talking about which subject they each studied hard.’

M: Swuhak-ul1 na-nun [VP yelsimhi t1 kongpwuha]-yss-e. Ne-nun? math-acc I-top hard study-past-decl you-top ‘I studied math hard. How about you?’

H: Swuhak-ul na-to kulay-ss-e. math-acc I-also so.do-past-decl Intended: ‘I studied math hard, too.’

(45) Simple-NoExtraction condition:

M: Na-nun [VP yelsimhi swuhak-ul kongpwuha]-yss-e. Ne-nun? I-top hard math-acc study-past-decl you-top ‘I studied math hard. How about you?’

H: Na-to kulay-ss-e. I-also do.so-past-decl Intended: ‘I studied math hard, too.’

(46) Complex-Extraction (long distance scrambling) condition:

Minswu-wa Hanswu-ka kakca etten kwamok-ul Yuli-ka yelsimhi Minswu-and Hanswu-nom each which subject-acc Yuli-nom hard kongpwuha-yess-tako tul-ess-nunci malha-ko iss-ess-ta. study-past-comp hear-past-comp talk-prog past-decl ‘Minswu and Hanswu were talking about which subject they each heard that Yuli studied hard.’

M: Swuhak-ul1 na-nun [VP[CP Yuli-ka yelsimhi t1 kongpwuha-yss-tako] math-acc I-top Yuli-nom hard study-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

91 ‘I heard that Yuli studied math hard. How about you?’

H: Swuhak-ul na-to kulay-ss-e. math-acc I-also do.so-past-decl

Intended: ‘I also heard that Yuli studied math hard.’

(47) Complex-NoExtraction condition:

‘Minswu and Hanswu were talking about which subject they each heard that Yuli studied hard.’

M: Na-nun [VP[CP Yuli-ka yelsimhi swuhak-ul kongpwuha-yss-tako] I-top Yuli-nom hard math-acc study-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

‘I heard that Yuli studied math hard. How about you?’

H: Na-to kulay-ss-e. I-also do.so-past-decl

Intended: ‘I also heard that Yuli studied math hard.’

In (44), the target sentence is assumed to have been derived by “short-scrambling” swuhak- ul ‘apple-acc’ out of kulay, which would correspond to the simple VP in the antecedent sentence, [VP yelsimhi t1 kongpwuha] ‘study t1 hard’. In (45), the target sentence does not involve scrambling, and kulay corresponds to the simple VP in the antecedent sentence, [VP yelsimhi swuhak-ul kongpwuha] ‘study math hard’. In (46), swuhak ‘apple-acc’ in the target sentence is assumed to have been “long-scrambled” out of kulay, which would correspond to the complex VP in the antecedent sentence, [VP [CP Yuli-ka yelsimhi t1 kongpwuha- yss-tako] tul] ‘hear that Yuli studied t1 hard’. In (47), the target sentence involves no scrambling, and kulay corresponds to the complex VP in the antecedent sentence [VP [CP Yuli-ka yelsimhi swuhak-ul kongpwuha-yss-tako] tul] ‘hear that Yuli studied math hard’. The primary conditions of interest were the Simple-Extraction and Complex-Extraction conditions, since the aim of the experiment was to investigate whether or not both short distance and long distance scrambling are available out of the VP anaphora. The Simple- NoExtraction and Complex-NoExtraction conditions, on the other hand, served as controls since the target sentences in these conditions are generally acceptable to Korean speakers. Following the pattern illustrated above, 24 sets of test items were constructed, generating 96 items (24 items for each of the four conditions). These items were then distributed across

92 four separate lists using a Latin Square design such that each list contained 24 test items, and each participant saw only one version of each item. The same 66 fillers were then added to each list. Crucially, 24 of the filler items had a similar design as the 24 test items in each list, except that their target sentences did not contain the VP anaphor kulay. These 24 filler items are termed “non-VPA control(s)” and used for subsequent statistical analysis. Sample sentences for these 24 filler items are given in (48)-(51) below, where the context sentences are given in English here for convenience and the target sentences are in boldface.

(48) Simple-Extraction (short distance scrambling) condition: Minswu and Hanswu were talking about which subject they each studied hard.

M: Swuhak-ul1 na-nun [VP yelsimhi t1 kongpwuha]-yss-e. Ne-nun? math-acc I-top hard study-past-decl you-top

‘I studied math hard. How about you?’

H: Swuhak-ul na-to [VP yelsimhi t1 kongpwuha]-yss-e. math-acc I-also hard study-past-decl

‘I studied math hard, too.’ (49) Simple-NoExtraction condition: Minswu and Hanswu were talking about which subject they each studied hard.

M: Na-nun [VP yelsimhi swuhak-ul kongpwuha]-yss-e. Ne-nun? I-top hard math-acc study-past-decl you-top

‘I studied math hard. How about you?’

H: Na-to [VP yelsimhi swuhak-ul kongpwuha]-yss-e. I-also hard math-acc study-past-decl

‘I studied Math hard, too.’ (50) Complex-Extraction (long distance scrambling) condition: Minswu and Hanswu were talking about which subject they each heard that Yuli studied hard.

M: Swuhak-ul1 na-nun [VP[CP Yuli-ka yelsimhi t1 kongpwuha-yss-tako] math-acc I-top Yuli-nom hard study-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

‘I heard that Yuli studied math hard. How about you?’

H: Swuhak-ul na-to [VP[CP Yuli-ka yelsimhi t1 kongpwuha-yss-tako] math-acc I-also Yuli-nom hard study-past-comp tul]-ess-e. hear-past-decl

93 ‘I also heard that Yuli studied math hard.’ (51) Complex-NoExtraction condition: Minswu and Hanswu were talking about which subject they each heard that Yuli studied hard.

M: Na-nun [VP[CP Yuli-ka yelsimhi swuhak-ul kongpwuha-yss-tako] I-top Yuli-nom hard math-acc study-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli studied math hard. How about you?’

H: Na-to [VP[CP Yuli-ka yelsimhi swuhak-ul kongpwuha-yss-tako] I-also Yuli-nom hard math-acc study-past-comp tul]-ess-e. hear-past-decl ‘I also heard that Yuli studied math hard.’

3.3.3.4 Procedure

The experiment was implemented using PsychoPy, as in Experiment 3. Each participant started the experiment with six practice trials, followed by 24 test trials (six trials per condition) and 66 ﬁllers in a uniquely generated random order. On average, each participant took 25-30 minutes to complete the experiment. They were paid $10 each as compensation for participation.

3.3.4 Findings

Figure 3.6 below summarizes mean acceptability ratings by condition: 1.71 in the Complex- Extraction (long distance scrambling) condition, 2.31 in the Simple-Extraction (short distance scrambling) condition, 5.63 in the Complex-NoExtraction condition, and 6.37 in the Simple-NoExtraction condition.30 Based on Sprouse’s (2016) recommendation on mapping grammaticality to a 7-point acceptability scale, it could be qualitatively said that while the target sentences for the two NoExtraction conditions were judged grammatical (> 5 mean rating), those for the two Extraction conditions were judged ungrammatical (< 3 mean rating). Contra Park (2015), this suggests that short distance as well as long distance scrambling are not available to Korean VP anaphora, which is further substantiated by the statistical analysis below. A linear mixed-effects analysis of the participants’ acceptability ratings, with antecedent VP type and extraction as fixed effects and participant and item as random effects, revealed a main effect of extraction (estimated coefficient = 3.92, se = .09, t = 41.89,

30“(+VPA)” is used in Figure 3.6 to indicate that the target sentences for all four test conditions include the VP anaphor kulay.

94 7 Complex Simple 6 5 4 Rating 3 2 1 0 Extraction(+VPA) NoExtraction(+VPA)

Figure 3.6: Mean acceptability ratings and standard errors for test conditions in Experiment 4 p < .001). This indicates that regardless of the antecedent VP type, speakers are likely to rate VP anaphora sentences with no scrambling higher than those with scrambling. In other words, sentences involving either short or long distance scrambling out of kulay are likely to be judged less acceptable than their corresponding non-scrambled sentences. Ac- cordingly, the extraction effect could, indeed, be interpreted as an indication that overt extraction out of the VP anaphora is hardly available in Korean. The analysis also revealed a main effect of antecedent VP type (estimated coefficient = .61, se = .09, t = 6.54, p < .001). That is, regardless of whether scrambling has been involved or not, speakers are likely to rate VP anaphora sentences with simple VP antecedents higher than those with complex VP antecedents. This was also substantiated by the pairwise comparisons of the mean ratings using a Tukey’s test, which revealed that not only is the rating in the Simple-Extraction condition significantly higher than the rating in the Complex-Extraction condition (p < .001), but also the rating in the Simple-NoExtraction is significantly higher than the rating in the Complex-NoExtraction condition (p < .001). Crucially, however, no interaction between Extraction and Antecedent VP type was found. These results thus suggest that VP anaphora sentences with short distance scrambling were rated higher than the sentences with long distance scrambling, not because the VP anaphor only allows for short distance scrambling, but because shorter and simpler antecedents are generally easier to comprehend than longer and more complex antecedents (Murphy 1985; cf. Frazier & Clifton 2001).

95 7 Complex Simple 6 5 4 Rating 3 2 1 0 Extraction(-VPA) NoExtraction(-VPA)

Figure 3.7: Mean acceptability ratings and standard errors for “non-VPA control” conditions in Experiment 4

There is a possibility that the effect of extraction observed above (or the low mean ratings in the Complex-Extraction and Simple-Extraction conditions) has merely stemmed from the general scrambling effect, which has often been reported to exist in the Korean language. Jackson (2008: 81), for instance, demonstrates in his experimental research that scrambled sentences in Korean are generally more difficult to understand (or process) than their corresponding non-scrambled counterparts, and that “the difficulty associated with scrambling can be reduced by contexts containing prior mention of the scrambled item”, but nevertheless does not completely disappear (see also Miyamoto 2003 and J. H. Kim et al. 2009 for a similar argument for scrambling in Japanese). Indeed, this scrambling effect was also borne out in the present experiment. That is, a linear mixed-effects analysis of the participants’ responses in the “non-VPA control” fillers revealed a main effect of extraction (estimated coefficient = 2.02, se = .09, t = 21.30, p < .001), as well as a main effect of antecedent VP type (estimated coefficient = .83, se = .09, t = 8.7, p < .001) (see Figure 3.7 above).31 Given these results, a further analysis was carried out to investigate whether the extraction effect observed in the VP anaphora sentences in the test conditions can be reduced to the general scrambling effect. To do so, we compared “VPA difference score(s)” to “NonVPA differences score(s)”. First, in the VP anaphora contexts, each participant’s VPA difference scores were calculated by subtracting her mean rating in the Complex-Extraction

31As opposed to “(+VPA)” in Figure 3.6, “(-VPA)” is used here to indicate that the target sentences for the four control conditions did not include the VP anaphor kulay.

96 condition from that in the Complex-NoExtraction condition; and by subtracting her mean rating in the Simple-Extraction condition from that in the Simple-NoExtraction condition. For instance, if a participant’s mean ratings for the Complex-Extraction and -NoExtraction conditions are 2 and 3, then her (Complex) VPA difference score is 1; also, if a participant’s mean ratings for the Simple-Extraction and -NoExtraction conditions are 1 and 6, then her (Simple) VPA difference score is 5. In a similar way, the NonVPA difference scores were calculated with the non-VPA control fillers. Now a prediction can be formulated as follows.

(52) prediction: If the effect of extraction in the VP anaphora contexts indeed falls out from the ‘general’ scrambling effect, then the mean VPA difference scores should not be significantly different from the mean NonVPA difference scores.

A linear mixed-effects analysis of the participant’s difference scores revealed that regardless of Complex versus Simple, the mean VPA difference scores were significantly higher than the mean NonVPA difference scores (estimated coefficient = 1.90, se = .20, t = 9.54, p < .001) (see Figure 3.8 below). This result does not cohere with the prediction given in (52), and thus it is now safe to say that the extraction effect in the VP anaphora contexts cannot be reduced to the general scrambling effect; rather, it must be because scrambling has been applied to the VP anaphora. These results taken together suggest that movement out of VP anaphora is not permissible in Korean. 5 Complex Simple 4 3 2 Difference score Difference 1 0 VPADifference NonVPADifference

Figure 3.8: Mean VPA diﬀerence and NonVPA diﬀerence scores and standard errors in Experiment 4

97 3.3.5 Discussion of Experiment 4

In addition to Experiment 3, the ﬁndings of Experiment 4 provide empirical evidence supporting the pro-form analysis of Korean VP anaphora. As discussed previously, if the VP anaphora were derived via deletion, it would, in principle, allow either short distance scrambling or long distance scrambling out of it, as illustrated in (53B) and (54B) (repeated from (40) and (41)), since it would contain enough structure to house the trace of a movement. Alternatively, if the deletion process were subject to the ‘anti-A¯ trace constraint’, as claimed by Park (2015), only the short distance scrambling example in (53B), not the long distance one in (54B), would be acceptable to Korean speakers. However, it has been demonstrated in Experiment 4 that neither short distance nor long distance scrambling is possible out of the VP anaphora. Our results can be readily accounted for if the VP anaphora is a pro-form, which is a syntactically atomic object from the outset of the derivation, so that there is no extractable phrase to begin with.

(53) A: Manhwa-lul1 na-nun [VP cacwu t1 ilk]-ess-e. Ne-nun? comic.book-acc I-top often read-past-decl you-top ‘I often read comic books. How about you?’

B: Manhwa-lul na-to kulay-ss-e. comic.book-acc I-also so.do-past-decl

Intended: ‘I often read comic books, too.’

(54) A: Manhwa-lul1 na-nun [VP [CP Yuli-ka cacwu t1 ilk-ess-tako] comic.book-acc I-top Yuli-nom often read-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli often read comic books. How about you?’

B: Manhwa-lul na-to kulay-ss-e. comic.book-acc I-also so.do-past-decl

Intended: ‘I also heard that Yuli often read comic books.’

Although our data does not support the short−long scrambling asymmetry observed in Park (2015) that short scrambling from the VP anaphora is grammatical but long scrambling is ungrammatical, we did find a statistically significant difference between the mean acceptability ratings for the Complex-Extraction (long distance scrambling) and Simple- Extraction (short distance scrambling) conditions (1.71 and 2.31).32 As noted earlier in

32Note that the mean ratings for the two scrambling conditions were as low as the mean ratings for the two ﬁller conditions (i) and (ii) below, both of which contained a target sentence that was expected to be (completely) unacceptable.

(i) Simple-“No maching” condition (mean rating: 1.75): Minho and Yengho were talking about what type of movie they often watched.

98 Subsection 3.3.4, however, this can reasonably be attributed to the main effect of antecedent vp type: in both scrambling and non-scrambling sentences, the ones with simple antecedent VPs were rated slightly higher than the ones with complex antecedent VPs. Also, even in non-VPA control sentences, simple sentences were rated slightly higher than complex sentences, regardless of scrambling, as shown in Figure 3.7. The current findings are, therefore, very likely to be reducible to the ease of comprehensibility or processing of simple clauses as opposed to more complex clauses. In light of our findings, the source of short-long scrambling asymmetry reported in Park (2015) may be processing/comprehensibility, or idiolectal/dialectal variation that might exist in Korean.33

3.4 General discussions

In this chapter, I have provided two bodies of novel empirical data that converge to support that Korean VP anaphora has no internal syntactic structure. In particular, I have found that Korean speakers unequivocally accept sloppy readings for the VP anaphora, irrespective of whether the pronoun ku can behave as a bound variable in their own grammar, and also that they are intolerant towards overt extraction out of the VP anaphora. Either of

M: Na-nun cacwu kongphoyenghwa-lul po-ass-e. I-top often scary.movie-acc watch-past-decl ‘I often watched scary movies.’

Y: Na-to cacwu kongphoyenghwa-lul an po-ass-e. I-also often scary.movie-acc neg watch-past-decl ‘I did not often watch scary movies, neither.’ (ii) Right scramble condition (mean rating: 1.95): Minswu and Hanswu were talking about which subject they each heard that Yuli studied hard. M: Na-nun Yuli-ka yelsimhi swuhak-ul kongpwuha-yss-tako tul-ess-e. I-top Yuli-nom hard math-acc study-past-comp hear-past-decl ‘I heard that Yuli studied math hard.’

H: Na-to Yuli-ka yelsimhi t1 kongpwuha-yss-tako swuhak-ul1 I-also Yuli-nom hard study-past-comp math-acc tul-ess-e. hear-past-decl Intended: ‘I also heard that Yuli studied math hard.’

In the target sentence in (ii), the embedded object swuhak-ul ‘math-acc’ has been “right-scrambled” over the embedded verb kongpwuha ‘study’. According to E. S. Lee (2008), a sentence with such scrambling is not acceptable in Korean. 33One difference between Park’s scrambling examples and ours is that while the scrambled constituents in Park were contrastive, ours were not. For instance, in Park’s example in (37a), say-lul ‘bird-acc’ in the antecedent clause contrasts with cwi-lul ‘mouse-acc’ in the VPA clause, while in our example in (44), the scrambled constituent in both the antecedent clause and the VPA clause is the same, swuhak-ul ‘math-acc’. Perhaps contrastive constituents in short scrambling (but not in long scrambling) have an ameliorating effect for some speakers, contributing to the reported enhanced acceptability of Park’s short scrambling examples. In any event, this does not affect our results since short and long non-contrastive scrambling were judged acceptable without kulay ‘do so’, but both short and long scrambling with the VP anaphor were judged unacceptable in comparison. I leave it to future work to determine whether contrastive scrambling interacts with the type of scrambling (short vs. long) and with VP anaphora.

99 these ﬁndings would be unexpected if the VP anaphora did indeed contain ‘silent’ structure. Drawing on these empirical ﬁndings, I argue that the VP anaphora in Korean does not result from the deletion process, but originates as a verbal pro-form, and is interpreted through semantic rules building an anaphoric dependency with semantic entities, e.g., events or situations, in a discourse. The pro-form analysis of Korean VP anaphora receives cross-linguistic support from its counterparts in other East Asian languages (e.g., Chinese and Japanese), which have been consistently analyzed as instances of pro-form anaphora in the existing literature. For example, in their recent extensive research on the syntactic structure and derivation of Mandarin Chinese VP anaphora zheme-zuo ‘do so’, Wei and Li (2016: 198) argue that the VP anaphor is “base-generated as it is, not derived by deleting a VP licensed by an antecedent”, using the impossibility of overt extraction as a supporting argument, as illustrated below.34

(55) Juzi1, wo xiwang ta manman chi t1; *pingguo, wo xiwang ta ye zheme orange I hope he slowly eat apple I hope he also so zuo. do Intended: ‘I hope that he slowly eats oranges; I hope that he also slowly eats apples.’ (adapted from Wei & Li 2016: 198, ex.(35))

The Japanese VP anaphor soo-su ‘do so’ has also been argued to be a pro-form (e.g., Hoji 1998, 2003; Fukaya & Hoji 1999). Observing that sloppy readings for pro-forms are characteristically diﬀerent from those under ellipsis, Hoji (2003) shows that soo-su ‘do so’ can induce a sloppy reading with an R-expression, as in (56), while instances of ellipsis phenomena in Japanese cannot.35

34Wei and Li (2016) also show that zheme-zuo ‘do so’ does not necessarily need a linguistic antecedent, as in (i), and can have split antecdents, as in (ii), either of which has been widely taken as an indication that an anaphoric object is a pro-form, not an instance of ellipsis, since Hankamer and Sag (1976) (cf. Merchant 2013a, 2013b).

(i) Observing John ripping a book in half: a. Ni Kebie zheme zuo! you do.not so do ‘You should not do so.’ (Wei & Li 2016: 196, ex.(29a)) b. Ni keneng zheme zuo ma? you possible so do Q ‘Will you possibly do so?’ (Wei & Li 2016: 196, ex.(29b)) (ii) Lisi keyi zhu fan, Zhangsan keyi xi cai, wo ye keyi zheme zuo, yiqi zhunbei. Lisi can cook meal Zhangsan can wash vegetable I also can so do together prepare ‘Lisi can cook meal; Zhangsan can wash vegetables; I can do so (=cook meal and wash vegetables) as well, (let’s) prepare together.’ (Wei & Li 2016: 196, ex.(30))

35Hoji (2003) demonstrates that in the Japanese “comparative ellipsis” construction in (i) below, the sloppy reading (ia) is marginally possible or simply impossible with the R-expression Bill within the CP.

100 (56) John-ga [VP John-no kuruma-o ara]-tta; Bill-mo soo sita. John-nom John-gen car-acc wash-past Bill-also so did ‘John washed John’s car; Bill did so, too.’

a. ‘John washed John’s car; Bill washed Bill’s car, too.’ [sloppy identity] b. ‘John washed John’s car; Bill washed John’s car, too.’ [strict identity] (Hoji 2003: 191, ex.(46))

Given that the proper name John cannot be semantically bound, soo su ‘do so’ could never involve the deletion of a syntactic constituent containing a bound variable element, which would otherwise be the source of the sloppy interpretation in (56a). It is thus reasonable to postulate that soo su ‘do so’ is a pro-form, and the sloppy reading is licensed via semantic/pragmatic inferences. Note that in line with Hoji (2003), both Bae and Kim (2012) and Park (2013) support the pro-form analysis of the Korean VP anaphor by demonstrating that it can also allow a sloppy reading with an R-expression, as illustrated in (57).36

(57) John-i [VP pi onun nal-e John-uy cha-lul takk]-ass-e. John-nom rain falling day-at John-gen car-acc wash]-past-decl Bill-to kulay-ss-e. Bill-also so.do-past-decl ‘John washed John’s car on a rainy day. Bill did so, too.’

a. ‘...Bill washed Bill’s car on a rainy day, too.’ [sloppy identity] b. ‘...Bill washed John’s car on a rainy day, too.’ [strict identity] (Bae & Kim 2010: 63, ex.(34))

Before concluding this chapter, there is one remaining issue that needs to be addressed. As stated previously, the ellipsis analysis is relatively more prevalent in the extant literature

(i) John-ni yorimo sakini sensei-wa Bill-ni [CP Mary-ga Bill-o butta to] iw-aseta. John-dat than earlier teacher-top Bill-dat Mary-nom Bill-acc hit com say-made ‘The teacher made Bill say that Mary had hit Bill earlier than John.’ a. ‘The teacher made John say that Mary had hit John.’ [*/*?sloppy identity] b. ‘The teacher made John say that Mary had hit Bill.’ [strict identity]

(Hoji 2003; 194, ex.(53))

36Following J. S. Kim’s (1997) proposal that the Korean focus construction involves VP ellipsis, Bae and Kim (2012) show that sloppy readings are not available with an R-expression in that construction, as illustrated in (i).

(i) John-i pi onun nal-e John-uy cha-lul takk-ass-e. Bill-to1 [VP t1 John-nom rain falling day-at John-gen car-acc wash]-past-se Bill-also pi onun nal-e John-uy cha-lul takk] ya. be ‘John washed John’s car on a rainy day. Bill (washed John’s/*Bill’s car on a rainy day), too.’ (Bae & Kim 2010: 62, ex.(32))

101 than the pro-form analysis. Notably, most of the proponents of the former analysis (e.g., Son 2006; Storoshenko 2008; Kim & Yoon 2009; Madigan 2015) uniformly rely on the well- known observation made by Cho (1996: 631) that VP anaphora sentences involving caki ‘self’ can only give rise to a sloppy reading, as illustrated in (58).

(58) John-i1 [VP caki-lul1 kwasinhay]-ss-ko, Mary-to kuleha-yess-ta. John-nom self-acc overtrust-past-conj Mary-also so.do-past-decl ‘John overtrusted himself, and Mary did, too.’ a. ‘John overtrusted John, and Mary overtrusted Mary.’ [sloppy identity] b. ‘John overtrusted John, and Mary overtrusted John.’ [*strict identity] (Cho 1996: 631, ex.(34)) Given the assumption that caki ‘self’ can (and should) only be construed as a bound variable, but not as a free variable (e.g., Han & Storoshenko 2012), the ellipsis analysis can easily account for the sloppy−strict asymmetry in (58) by positing that caki is present within the site of kuleha, with its intrinsic property intact, so that it cannot be coreferential with the subject of the first clause (i.e., John), and can only be bound to the clausemate subject (i.e., Mary). Under the pro-form analysis, the unavailability of the strict reading seems to be problematic because it seems difficult to provide a principled reason to prevent the VP anaphor from referring to the “overtrusting John” event, which could plausibly be inferred from the antecedent VP. It should be noted here, however, that the unavailability of the strict reading is subject to pragmatic influence. For example, the strict reading in (58b) becomes readily available if the sentence is presented in a pragmatic context favouring the strict reading, as in (59).

(59) John and Mary are high school students. John is very good at biology, but Mary is not really good at it. They were studying together at the library for a biology exam the next day. Unlike John, Mary was not really ready for it. John wanted to help her, so he said, “Hey, I am pretty sure about what questions will be asked in the exam. I will explain them to you now, so you only focus on them, ok? Trust me, you will get at least a B+.” “Thanks a lot! I trust you, John,” said Mary. On the next day, they took the biology exam. It turned out that most of the questions covered in the exam were not relevant to what they studied the previous day. Eventually, they both ended up getting a C−.

Also, even in the absence of context, the sentence in (60) below can easily have the strict reading in (60b), where the subject of the second conjunct is lower in the social hierarchy than the subject of the ﬁrst conjunct, according to general pragmatic knowledge.

(60) Leehwuechang-i1 [VP caki-lul1 kwasinhay]-ss-ko, ku-uy pise-to Lee.president-nom self-acc overtrust-past-conj he-gen secretary-also kuleha-yess-ta. so.do-past-decl

102 ‘President Lee overtrusted himself, and his secretary did so, too.’ a. ‘..., and his secretary overtrusted himself, too.’ [sloppy identity] b. ‘..., and his secretary overtrusted President Lee, too.’ [strict identity]

The above observations suggest that, in principle, the VP anaphora sentences involving caki ‘self’ can allow strict readings as well as sloppy readings. This being so, the sloppy−strict asymmetry in (58) may be taken merely as an indication that the strict reading is not preferred over the sloppy reading, therefore not actually posing any problem to the pro- form analysis of Korean VP anaphora.

103 Chapter 4

The syntax of null objects in Korean

4.1 Introduction

One of the deﬁning grammatical properties of East Asian languages such as Korean, Japanese, and Chinese is that they freely allow the omission of object arguments (as well as subject arguments).1 This object drop phenomenon can be attested in the so-called null object construction, as exempliﬁed by the Korean sentences in (1) and (2), where a transitive verb with no overtly expressed object (marked with [e] as a theory-neutral notation) appears in the second conjunct of each coordinate structure.

(1) Sengkyeng-ey po-myen, Hawa-ka mence senakkwa yelmay-lul mek-ko, Bible-at see-if Hawa-nom ﬁrst good.evil.fruit fruit-acc eat-conj ku hwu-ey Adam-to [e] mek-supni-ta. that after-at Adam-also eat-hon-decl (lit.) ‘According to the Bible, Hawa(=Eve) ﬁrst eats the fruit of knowledge of good and evil, and then Adam eats, too.’ (https://www.leesangku.org/ns/?mid=board_iKrY97&page=26&document) (2) Appa-nun sayenni-lul coaha-yss-ciman, emma-nun [e] sileha-yss-ta. dad-top new.sister-acc like-past-conj mom-top dislike-past-decl (lit.) ‘Dad liked older brother’s wife, but mom disliked.’ (https://torantoran.postype.com/post/763142)

In the second conjunct in (1), the object of mek ‘eat’ (as a transitive verb) is missing, but it is readily understood as corresponding to the phonologically overt object in the ﬁrst

1Earlier versions of this chapter were presented at the 29th Annual CUNY Conference on Human Sentence Processing held at the University of Florida, Gainesville, in April 2016, the Workshop on Experimental Approaches to East Asian Languages held at the University of Hawai‘i at M¯anoa,in May 2016, and the 24th Japanese/Korean Linguistics Conference at NINJAL, Japan, in October 2016.

104 conjunct, senakkwa yelmay ‘the fruit of knowledge of good and evil’. Likewise, in the second conjunct in (2), the null object of sileha ‘dislike’ is immediately taken to denote the same referent as the audible counterpart in the ﬁrst conjunct, sayenni ‘older brother’s wife’. This is in contrast to the English coordinate structure in (3) (the literal translation of (1)), which demonstrates that object omission is not grammatical in English.2

(3)* Hawa ﬁrst eats the fruit of knowledge of good and evil, and then Adam eats [e], too.

Note that the second conjunct in (3) can be made felicitous by either constructing a VP ellipsis instead, as in (4a), or replacing [e] with a full nominal expression or a pronoun it, as in (4b).

(4) a. ..., and then Adam does [VP eat the fruit of knowledge of good and evil], too. b. ..., and then Adam eats {the fruit of knowledge of good and evil/it}, too.

The precise characterization of the syntactic status of null complement objects has been one of the extensively debated issues in the literature of East Asian languages over the past three decades or so. As suggested by Ahn and Cho (2011a), Funakoshi (2016), and Sato and Karimi (2016), the existing proposals can be grouped into at least three camps: (i) the null pronominal analysis; (ii) the argument ellipsis analysis; and (iii) the verb-stranding VP ellipsis analysis.3

2As attested in Wilder’s (1997: 74) example in (i), along with (3) above, it has generally been assumed that English is a non-null object (and non-null subject) language (e.g., Roberts & Holmberg 2010). Note, however, that English might employ object dropping in certain limited written contexts, such as recipes (Massam & Roberge 1989, Massam 1992), as in (ii), and diary-style texts (Haegeman 1990, Haegeman & Ihsane 1999), as in (iii).

(i)* John bought the book, and Mary read [e]. (Wilder 1997: 74, ex.(52)) (ii) Take the cake mix, 1 cup of water, and 3 eggs. Mix [e] well and beat [e] for 5 minutes. Pour [e] into a well-greased cake pan and bake [e] for 20 minutes. Remove [e] from oven and cool [e]. (Massam & Roberge 1989: 135, ex.(2)) (iii) Search for hairbrush. Locate [e] in handbag. (Haegeman & Ihsane 1999: 100, ex.(2))

See Massam and Roberge (1989: 138) for an argument that the English null objects might be analyzed as either a variable bound by a null topic (see Footnote 3 below) or “a sort of NP-trace (a trace bound by an element in an A-position)”. 3An alternative account for null objects is the null topic variable analysis, which was proposed in Huang (1982, 1984a, 1984b) and Hasegawa (1984/85). According to this analysis, languages like Chinese and Japanese can license a null topic in an internal argument position, which is subsequently A0-moved to the sentence-initial topic position. Consider ﬁrst the examples of Chinese null objects and subjects below.

(i) [Zhangsan2 xiwang [Lisi keyi kanjian [e]]]. Zhangsan hopes Lisi can see

(lit.) ‘Zhangsan2 hopes Lisi can see [e](=him1/∗2).’ (compare: Zhangsan2 xiwang Lisi keyi kanjian ta. ‘Zhangsan2 hopes Lisi can see him1/2.’

(ii) [Zhangsan2 xiwang [[e] keyi kanjian Lisi]]. Zhangsan hopes can see Lisi

105 It has been advocated by a number of researchers that a phonologically null pronoun (pro in Chomsky’s (1981, 1982) typology) is base-generated in the null object position [e] (e.g., Yoon 1985; Cole 1987; M. K. Park 1994, 1997; Hoji 1998, 2003; G. Li 1998; S. H. Kim 2010; Moon 2010; Ahn & Cho 2010, 2011a, 2011b; Bae & Kim 2012), as illustrated in (5), where irrelevant structural details are abstracted away and English translation is used for convenience.

(lit.) ‘Zhangsan2 hopes [e](=he1/2) can see Lisi.’ (compare: Zhangsan2 xiwang ta keyi kanjian Lisi. ‘Zhangsan2 hopes he1/2 can see Lisi.’ (adapted from Huang 1984a: 538, ex.(22a) and (22b))

In (i), the null object can only correspond to an entity introduced in the preceding context, but not to a clause-internal argument, such as the matrix subject Zhangsan, while the null subject does not exhibit such restricted interpretive properties, as shown in (ii). Huang argues that the asymmetry is to be expected if the null object and subject have diﬀerent syntactic statuses, as represented below.

(iii) [OP1 Zhangsan2 xiwang [Lisi keyi kanjian t1]].

(iv) [Zhangsan2 xiwang [pro keyi kanjian Lisi]].

In (iii), the null object is a trace that is locally A0-bound by the null topic operator and thus functions as a variable. It would be then subject to Principle C, just like an R-expression, thus obeying the Strong Crossover Constraint (Chomsky 1981, 1982). In (iv), on the other hand, the null subject position is occupied by an empty pronoun, pro, which would be subject to Principle B, not Principle C. Under this view, then, the Korean null object construction in (1) would be represented as in (v), where the ‘null variable object’ is bound by the null topic, which in turn refers to the previously mentioned object, senakkwa yelmay ‘the fruit of knowledge of good and evil’.

(v) Hawa-ka mence [DP senakkwa yelmay-lul] mek-ko, ku hwu-ey [OP1 Adam-to t1 mek-supni-ta].

As noted by Cole (1987: 602), however, null objects in Korean should not be treated as variables, since they can naturally be anaphoric to “matrix arguments in unmarked contexts”, as illustrated in (vi)-(viii) below.

(vi) Ku yeca1-ka [namphyen-i [e] phokhayngha-yess-ta-ko] kecitmalha-yess-ta. that woman-nom husband-nom hit-past-decl-comp lie-past-decl

(lit.) ‘That woman1 lied that her1 husband hit [e](=her1/2).’ (Google)

(vii) Chelswu1-ka [Yenghi-ka [e] hyeppakha-yess-ta-ko] cwucangha-yess-ta. Chelswu-nom Yenghi-nom threaten-past-decl-comp claim-past-decl

(lit.) ‘Chelswu1 claimed that Yenghi threatened [e](=him1/2).’

(viii) John1-un [Bill-i [e] cenhwaha-yess-ta]-nun sasil-ul acik moru-n-ta. John-top Bili-nom call-past-decl-adn fact-acc yet not.know-pres-decl

(lit.) ‘John1 doesn’t know the fact that Bill called [e](=him1/2).’ (Cole 1987: 602, ex.(18a) and (18b), originally from Yoon 1985)

Given the above data, the null topic variable analysis does not seem to be empirically appropriate for Korean null objects, and thus is not considered as an analytic option in the current research.

106 (5) Null pronominal:

VP VP

DP V DP(=[e]) V

good.evil...-acc eat pro eat

Similar to the behaviour of a regular overt pronoun such as kukes ‘it’, the empty pronominal might refer to the (most) contextually salient entity, e.g., the fruit of knowledge of good and evil in (1), which has been introduced into the discourse by the overt object in the ﬁrst clause.4 An alternative view of the status of null objects is that they are the result of an ellipsis operation called argument ellipsis (or NP/DP ellipsis) (e.g., Oku 1998; S. W. Kim 1999; Saito 2003, 2007; Saito & An 2010; Takahashi 2007, 2008, 2010; Takita 2011; Um 2011; Cheng 2011, 2013; J. S. Kim 2012; M. K. Park 2013; Ohtaki 2011, 2014; Sakamoto 2015, 2016).5 That is, a full-ﬂedged DP constituent is normally constructed in the null object position [e], but may subsequently become unpronounced under identity with an overt DP in the corresponding object position in a preceding clause, as illustrated in (6).6

4In later sections, the term null pronominal analysis is used to indicate Hoji’s (1998) ‘indeﬁnite pro’ analysis (see Subsection 4.2.2 for further details). 5Argument ellipsis has also been argued to occur in South Asian languages such as Bangla, Hindi, and Malayalam (Simpson et al. 2013; cf. Takahashi 2013), and in other languages such as Colloquial Singapore English (Sato 2014, 2016), Mongolian (Takahashi 2007), Persian (Sato & Karimi 2016), and Turkish (Sener & Takahashi 2010). 6As pointed out by Takita (2011) and Takahashi (2013), the term argument ellipsis has been used, in part, to stress that adjuncts alone cannot be elided, as illustrated by the following Korean examples.

(i)A: Minswu-ka sinsokhi ku mwuncey-lul phwul-ess-ta. Minswu-nom quickly the problem-acc solve-past-decl ‘Minswu solved the problem quickly.’ B:* Cinho-nun [e] ku mwuncey-lul an phwul-ess-ta. Cinho-top the problem-acc neg solve-past-decl Intended: ‘Cinho did not solve the problem quickly.’ B0: Cinho-nun sinsokhi ku mwuncey-lul an phwul-ess-ta. Cinho-top quickly the problem-acc neg solve-past-decl (adapted from Takahashi 2013: 4, ex.(5)))

In (iB), the adverb sinsokhi ‘quickly’ is intended to have been elided under identity with the overt adverb in (iA). If ‘adjunct ellipsis’ indeed were a licit process, then the interpretation of (iB) should be ‘Cinho solved the problem, but he did not do so quickly’ just like that of (iB0), where the target adverb is explicitly expressed. However, (iB) is unanimously interpreted as ‘Cinho did not solve the problem at all’, thus suggesting that unlike arguments, adjuncts on their own cannot directly undergo ellipsis (see, however, Takahashi 2007, who argues that ‘adjunct ellipsis’ may be operative in Mongolian and Mandarin Chinese).

107 (6) Argument ellipsis:

VP VP

DP V DP(=[e]) V

good.evil...-acc eat good.evil...-acc eat

Another analytic option for null objects involves VP ellipsis preceded by overt movement of the main verb out of the VP to Tense (Huang 1987, 1988, 1991; Otani & Whitman 1991; cf. Pan 2002; E. J. Lee 2005b; Funakoshi 2014, 2016; Fujiwara 2017; cf. W. S. Lee 2016). In (7), for instance, all the elements within the VP in the second conjunct (i.e., the DP object and the trace left behind by V-raising) undergo ellipsis, whereas the verb in Tense, mek ‘eat’, is stranded outside the elided VP and is thus overtly realized. Therefore, under this view, the null object site [e] is actually equivalent to an articulated VP constituent, although it does not appear to be so on the surface.7

(7) Verb-stranding VP ellipsis:

TP TP

Hawa-nom T0 Adam-also T0

VP T VP(=[e]) T

DP V eatv DP V eatv

good.evil...-acc tv good.evil...-acc tv

Given the above considerations, the three competing approaches might be distinguished from one another in their assumptions regarding the following two parameters: (i) whether null objects contain unpronounced internal structure; and (ii) whether null objects correspond to DPs or VPs, as illustrated in Table 4.1 below.

Source of null objects (i) Internal structure? (ii) [e] is DP or VP? Null pronominal Absent DP Argument ellipsis Present DP Verb-stranding VP ellipsis Present VP

Table 4.1: Syntactic status of null objects

7The verb-stranding VP ellipsis analysis has also been proposed for Hebrew (Doron 1999; Goldberg 2005), Irish (McCloskey 1991), Persian (Shaﬁei 2015), and Russian (Gribanova 2013).

108 The primary purpose of this chapter is to present novel empirical data obtained from an experimental study (Experiment 5) to defend the position that some instances of Korean null objects should be analyzed as being derived via an ellipsis process. In Experiment 5, the (un)availability of sloppy identity readings for null objects was examined in order to diagnose the presence/absence of internal syntactic structure within them, in much the same way as in the investigation of the syntax of Korean VP anaphora in Experiment 3 in Chapter 3. I demonstrate that the results from Experiment 5 could only be obtained if the null object constructions tested are attributed to the ellipsis of a constituent with full-ﬂedged structure, thus undermining the view that Korean null objects are all instances of empty pronominals, atomic elements that lack internal structure (e.g., Ahn & Cho 2010, 2011a, b). This chapter does not aim to determine the ‘size’ of ellipsis involved in Korean null objects, i.e., whether the element elided in the position of [e] is a DP object (Argument ellipsis) or a VP containing the DP object (Verb-stranding VP ellipsis). However, Section 4.4 discusses a possible direction for future research on this issue, along with a brief review of the relevant literature.

4.2 Theoretical background

4.2.1 Korean as a radical pro-drop language

As noted in Section 2.2 of Chapter 2, it has long been agreed in the theoretical literature that East Asian languages, such as Korean, Japanese, and Chinese, and Romance languages, such as Spanish and Italian, are all characterized as pro-drop languages (or null subject languages), given the common characteristic that they allow an understood pronominal subject of a tensed clause to be ‘dropped’. Consider ﬁrst the Spanish sentences in (8) and (9).

(8) a. José sabe [que (él) ha sido visto por Maria]. José know that he have.subj.3sg been seen by Maria

‘José1 knows that he1/2 has been seen by Maria.’ b. Jose sabe [que Maria *(lo) ha visto]. José know that Maria him have.subj.3sg seen

‘José1 knows that Maria has seen him1/2.’ (adapted from Huang 1984a: 533, ex.(5) and (6)) (9)A: El hombre golpeó la mujer? the.masc.sg man hit.subj.3sg the.fem.sg woman ‘Did the man hit the woman?’ B: Sí, (él) *(la) golpeó. yes he her hit.subj.3sg ‘Yes, he hit her.’

109 Note above that the third-person singular subject pronoun, él ‘he’, can be omitted in the ﬁnite embedded and matrix clauses, whereas the object (clitic) pronoun, lo ‘him’, must be overtly realized. Such subject-object asymmetry is not attested in non-pro-drop (or non-null subject) languages such as English, where neither the pronominal subjects nor the pronominal objects of tensed clauses can be unexpressed, as exempliﬁed in (10) and (11).

(10) a. John promised Bill that [*(he) would see Mary]. b. John promised Bill that [Mary would see *(him)]. (adapted from Huang 1984a: 532, ex.(3a) and (3b)) (11)A: Did the man meet the woman? B: Yes, *(he) met *(her).

In the standard theory, the above cross-linguistic variation and language-internal variation with respect to the (un)availability of pro-drop are attributed to the presence or absence of ‘rich’ (or ‘uniform’) agreement under T within individual languages (Perlmutter 1971; Taraldsen 1978; Chomsky 1981, 1982; Rizzi 1982, 1986; Jaeggli & Safir 1989; Koeneman 2000, among many others). The basic idea goes as follows: Spanish (and Italian)-type languages have ‘rich’ verbal inflectional morphology for subject-verb agreement (e.g., the subject él ‘he’ and the verb golpeó ‘hit.subj.3sg’ in (9)), but none for object-verb agreement; thus, pro-drop may be licensed in a finite subject, but not object, position, since the content of an omitted pronoun could be readily ‘recoverable’ by means of the morpho-syntactic agreement. English-type languages, on the other hand, exhibit ‘impoverished’ subject-verb agreement morphology and no object-verb agreement morphology; thus, neither pronominal subjects nor pronominal objects may be omitted in these languages, since there would be no means to ‘recover’ the content of an otherwise missing pronoun.8 However, as already noticed by many researchers, including Chomsky (1981: 284) and Neeleman and Szendroi (2007: 647), the agreement-based theory of pro-drop, established mainly based on European languages, would be immediately challenged when considering East Asian languages such as Korean, Japanese, and Chinese. These languages typically do not have (overt) verbal agreement morphology at all, and thus would be predicted to exhibit total absence of pro-drop, which is not the case, however, as exemplified in (12)-(14) below. 8Agreement in English, for instance, is considered to be ‘impoverished’, given that it only displays one subject-verb agreement suffix -(e)s, which occurs with a third-person singular subject in present tense, as illustrated below.

(i) He/She likes/*like/liked playing soccer in the morning. (ii) They/You/I *likes/like/liked playing soccer in the morning.

110 (12) Korean

Kiho-ka Yuli-eykey [(ku-ka) (kunye-lul) kongwon-eyse po-ass-ta-ko] Kiho-nom Yuli-to he-nom she-acc park-at see-past-decl-comp malha-yess-ta. say-past-decl

‘Kiho said to Yuli that he saw her at the park.’ (13) Japanese

Yusuke-ga Hanako-ni [(kare-ga) (kanozyo-o) koen-de mita-to] itta. Yusuke-nom Hanako-to he-nom she-acc park-at saw-comp said

‘Yusuke said to Hanako that he saw her at the park.’ (14) Chinese

Zhangsan dui Fei shuo [(ta) zai gongyuan kanjian (ta) le]. Zhangsan to Fei said he at park saw her le

‘Zhangsan said to Fei that he saw her at the park.’

As can be seen above, Korean, Japanese, and Chinese do not inﬂect verbs for agreement, but nevertheless allow free omission of subject pronouns. Additionally, as Neeleman and Szendroi (2007: 647) put it, “pro-drop in [these ‘agreementless’ languages] seems to be more widespread than in [rich-agreement languages]”, since the pro-drop phenomenon takes place not only in the subject position, but also in all other argument positions, including direct/indirect object positions (see also Kuroda 1965, Huang 1982, 1984, Hoji 1985, Saito 1985).9 Given this ‘liberal’ pattern of pro-drop, Korean, Japanese, and Chinese have frequently been referred to as ‘radical pro-drop’ languages in the literature. Also, note that these East Asian languages are generally assumed to depend primarily on discourse to ‘recover’ dropped pronouns, especially as demonstrated by the Korean examples in (15), where either subject pro-drop or object pro-drop arises in contexts without a linguistic antecedent.10

(15) a. (Cinswu’s mother is pointing to a bruise on his forehead. Cinswu says to her:)

9As noted in Chapter 2, the Korean-type languages are also known to allow omission of pronouns in the possessor positions, as shown in the Korean example below. (i) Minswu-nun [(ku-uy) tongsayng-ul] cengmal salangha-n-ta. Minswu-top he-gen younger.brother-acc really love-pres-decl ‘Minswu really loves his younger brother.’

See also, e.g., Neeleman and Szendroi (2007: 647), for Japanese and Chinese examples. 10The exact characterization of licensing conditions for radical pro-drop is still an issue of considerable discussion. See Tomioka (2003), Speas (2004), Saito (2007), and Neeleman and Szendroi (2007) for detailed discussion.

111 Minswu-ka (na(-lul)) ttayli-ess-e. Minswu-nom I-acc hit-past-decl ‘Minswu hit me.’ b. (Cinswu’s mother has found that there is only 50 cent in Minswu(=Cinswu’s younger brother)’s wallet. Cinswu says to her:) (kyay(-ka)) cemsim-ey haympeke sey-kay(-lul) sa mek-ess-e. he-nom lunch-at hamburger three-cl-acc buy eat-past-decl ‘He bought and ate three hamburgers for lunch.’ c. (Cinswu has found Tongswu’s garage empty where there used to be his truck. Minswu says to Cinswu:) (kyay(-ka)) (kukes(-ul)) ecey phal-ass-e. he-nom it-acc yesterday sell-past-decl ‘He sold it yesterday.’ (adapted from Lee & Kim 2010: 1033)

Crucially, within the framework of generative grammar, a standard syntactic assumption about the cross-linguistic pro-drop phenomenon has been that a phonologically empty (but syntactically and semantically active) pronoun, the little pro, is licensed in the position of a dropped pronominal argument. Under this view, the Korean sentence in (15a) would be represented as in (16), where the object position is occupied by a pro, whose referent would be deduced from discourse.

(16) (Cinswu’s mother is pointing to a bruise on his forehead. Cinswu says to her:)

Minswu-ka pro ttayli-ess-e. Minswu-nom hit-past-decl ‘Minswu hit me.’

Since the work of Chomsky (1981, 1982), it has been customary to treat a pro as the null counterpart of a regular overt pronoun (e.g., ku ‘he’), having the features [−anaphor, +pronominal].11 Therefore, as discussed by, for example, Takahashi (2008: 308) and Fu- nakoshi (2016:114), a supporting argument for the occurrence of pro within an individual

11As has been observed in Chapter 2, however, within pro-drop languages, null and overt pronouns are not entirely in free variation. Recall that in Spanish and Italian, for example, null pronouns may be construed as bound variables while their overt counterparts may not. (i) Spanish Nadie cree que {pro/*él} es inteligente. nobody believe.pres comp he be.pres intelligent

Intended: ‘Nobody1 believes that he1 is intelligent.’ (ii) Italian Ogni studente crede che {pro/*lui} è intelligente. every student think.pres comp he be.pres intelligent

112 pro-drop language would be that it complies with Principle B of the binding theory on a par with an overt pronoun. This point is illustrated by the Korean examples in (17) and (18) below, which are all ungrammatical under the interpretation where the empty category represented by [e] and the overt pronoun ku ‘he’ are anaphorically connected to the subjects within the same clause.

(17) a.* Minho-ka simhakey [e] pinanha-yss-e. Minho-nom severely blame-past-decl Intended: ‘Minho blamed himself severely.’ b.* Motun haksayng-i simhakey [e] pinanha-yss-e. every student-nom severely blame-past-decl Intended: ‘Every student blamed himself severely.’ (18) a.* Minho-ka simhakey ku-lul pinanha-yss-e. Minho-nom severely he-acc blame-past-decl Intended: ‘Minho blamed himself severely.’ b.* Motun haksayng-i simhakey ku-lul pinanha-yss-e. every student-nom severely he-acc blame-past-decl Intended: ‘Every student blamed himself severely.’ (constructed on the basis of Takahashi’s (2008: 308) Japanese examples)

The sentences in (18) are ruled out due to a violation of Principle B, since the pronoun ku ‘he’ is not free in its local binding domain (e.g., Hong 1985). The ungrammaticality of the sentences in (17) can also be straightforwardly accounted for in terms of Principle B, if the null element in the object position is a pro, “a pure pronominal like its overt counterpart” (Chomsky 1982: 81-82).12 In light of the above considerations, it seems quite reasonable to conclude that employing pro in an argument position is a grammatical option that is independently available in Korean and other East Asian languages.13 Notably, however, there has been a considerable debate in the literature as to whether the pro strategy is the only way of deriving null arguments in these radical pro-drop languages. Some researchers have argued that there are cases of null objects that certainly do not involve pro, but must be the result of an ellipsis process (Huang 1987, 1988, 1991; Otani & Whitman 1991; Oku 1998; S. W. Kim 1999, among many others). However, this claim has been challenged by other researchers such as Hoji (1998, 2003), G. Li (1998), Ahn and Cho (2010, 2011a, 2011b), Moon (2010), and Bae and Kim (2012), who have proposed that the relevant null object examples can all

Intended: ‘Every student1 thinks that he1 is intelligent.’

12See, however, footnote 5 for Huang’s (1982, 1984a, 1984b) argument that an object empty category cannot be a null pronominal. 13See Takahashi (2008: 308), Ohtaki (2011: 248), and Funakoshi (2016: 114) for a similar discussion for Japanese.

113 be accommodated in terms of pro (more precisely, an ‘indeﬁnite pro’). In the next section, we further delve into this ellipsis versus null pronominal debate.

4.2.2 Ellipsis versus null pronominal

The ﬁrst and most frequently cited argument in defence of the ellipsis strategy for null objects in East Asian languages is that sloppy identity readings are available in some null object constructions, as illustrated in the Korean examples in (19) and (20) below.14 Note that the antecedent sentences in (19A) and (20A) have a DP object containing the long- distance anaphor caki ‘self’ as a possessor.

(19)A: Minswu 1-ka caki1-uy emma-lul piphanha-yess-ta. Minswu-nom self-gen mom-acc critisize-past-decl

‘Minswu1 criticized his1 mom.’ B: Cinswu-to [e] piphanha-yess-ta. Cinswu-also critisize-past-decl (lit.) ‘Cinswu criticized, too.’

a. ‘Cinswu criticized his own mom, too.’ [sloppy identity]

b. ‘Cinswu criticized Minswu’s mom, too.’ [strict identity]

(adapted from Ahn & Cho 2011: 473, ex.(3))15

(20)A: John 1-i caki1-uy khemphwute-lul pwuswu-ess-ta. John-nom self-gen computer-acc destroy-past-decl

‘John1 destroyed his1 computer.’

14See also Japanese and Chinese examples from S. W. Kim (1999: 255) below.

(i)A: Bill 1-wa zibun1-no tegami-o suteta. Bill-top self-gen letter-acc discarded

‘Bill1 discarded his1 letter.’ B: John-mo [e] suteta. John-also discarded (lit.) ‘John discarded, too.’ a. ‘John discarded his own letter, too.’ [sloppy identity] b. ‘John discarded Bill’s letter, too.’ [strict identity]

(ii)A: Zhangsan 1 bu xihuan guanyü ziji1-de yaoyan. Zhangsan not like about self-gen rumor ‘Zhangsan doesn’t like rumors about himself.’ B: Mali ye bu xihuan [e]. Mary also not like (lit.) ‘Mary doesn’t like, either.’ a. ‘Mary doesn’t like rumors about herself, either.’ [sloppy identity] b. ‘Mary doesn’t like rumors about Zhangsan, either.’ [strict identity]

15Ahn and Cho’s original example contains casin ‘self’, another Korean long-distance anaphor.

114 B: Peter-to [e] pwuswu-ess-ta. Peter-also destroy-past-decl (lit.) ‘Peter destroyed, too.’

a. ‘Peter destroyed his own computer, too.’ [sloppy identity]

b. ‘Peter destroyed John’s computer, too.’ [strict identity]

(adapted from Saito & An 2010: 3, ex.(8))

As indicated above, the null object sentences in (19B) and (20B) allow a sloppy identity reading as well as a strict identity reading. This interpretive pattern would pose a non- trivial challenge to the “uniform pro-theory” (Sato & Karimi 2016: 2). Recall that pro is standardly considered to be the null analogue of an overt pronoun. Therefore, if the null object arguments in (19B) and (20B) were simply empty pronominals, the sentences, contrary to fact, should not be able to yield the sloppy identity readings, on a par with the sentences in (21B) and (22B) below, which contain kunye ‘she’ and kukes ‘it’ in the object positions. Here, these overt pronouns can only be interpreted referentially and, thus, only allow the strict identity readings.

(21)A: Minswu 1-ka [caki1-uy emeni]2-lul piphanha-yess-ta. Minswu-nom self-gen mother-acc critisize-past-decl

‘Minswu1 criticized [his1 mother]2.’

B: Cinswu-to kunye2-lul piphanha-yess-ta. Cinswu-also her-acc criticize-past-decl

‘Cinswu criticized her2, too.’ [*sloppy identity, strict identity]

(22)A: John 1-i [caki1-uy khemphwute]2-lul pwuswu-ess-ta. John-nom self-gen computer-acc destroy-past-decl

‘John1 destroyed [his1 computer]2.’

B: Peter-to kukes2-ul pwuswu-ess-ta. Peter-also it-acc destroy-past-decl

‘Peter destroyed it2, too.’ [*sloppy identity, strict identity]

In order to accommodate the availability of sloppy interpretation in null object constructions like (19) and (20), a number of syntacticians have argued that the relevant null objects should be analyzed as resulting from some kind of ellipsis operation, which might be either argument ellipsis (e.g., Oku 1998; Kim 1999; Saito 2003, 2007; Takahashi 2007, 2008, 2010; Sakamoto 2015, 2017) or verb-stranding VP ellipsis (e.g., Huang 1987, 1991; Otani & Whitman 1991; E. J. Lee 2005; Funakoshi 2014, 2016), as previously introduced in Section 1. According to these ellipsis views, the null object construction in (19) may be represented in two ways, as illustrated in (23) and (24).

115 (23) Argument ellipsis:

VP VP

DP V DP(=[e]) V

caki-gen mother-acc criticize caki-gen mother-acc criticize (24) Verb-stranding VP Ellipsis:

TP TP

Minswu-nom T0 Cinswu-also T0

VP T VP(=[e]) T

DP V criticizev DP V criticizev

caki-gen mother-acc tv caki-gen mother-acc tv

In (23) and (24), either of the structures for the null object sentence in (19B) has the object DP embedding the possessive anaphor caki-uy ‘self-gen’, which is elided under identity with its overt antecedent. Therefore, the sloppy identity reading can be readily attributed to the elided anaphor being bound by the clause-mate subject, Cinswu. Notably, this ellipsis analysis can be further supported by the English data below. The sentence in (25a) involves an elided (VP) structure hosting a possessive pronoun his, which is available to serve as a bound variable and can thus induce a sloppy identity reading; the sentence in (25b), on the other hand, includes a referential pronoun her in the object position, and thus allows only a strict identity reading, but not a sloppy identity reading.

(25) a. Minswu1 criticized his1 mother, and Cinswu did [criticize his mother], too. [sloppy identity, strict identity]

b. Minswu1 criticized his1 mother, and Cinswu criticized her, too. [*sloppy identity, strict identity]

Another representative argument made to substantiate the ellipsis analysis is based on the observation that some null objects yield the ‘quantificational readings’ (the term coined by Takahashi 2008a, 2008b). Consider the following Korean null object construction, in which the antecedent sentence in (26A) contains a nominal object modified by a numerical quantified expression sey myeng(-uy) ‘three cl(-gen)’.16

16See also Japanese and Chinese examples from Sener and Takahashi (2011: 81-82) and Cheng (2014: 2) below.

116 (26)A: Minswu-ka sey myeng-uy kaswu-lul coaha-n-ta. Minswu-nom three cl-gen singer-acc like-pres-decl ‘Minswu likes three singers.’ B: Cinswu-to [e] coaha-n-ta. Cinswu-also like-pres-decl (lit.) ‘Cinswu likes, too.’

a. ‘Cinswu likes three singers, too.’ [quantiﬁcational]

b. ‘Cinswu also likes the same three singers who Minswu likes.’ [E-type]

As indicated above, the null object sentence in (26B) may be interpreted in two different ways. One reading is the quantificational reading, in which the set of singers that Cinswu likes do not need to be identical to the ones that Minswu likes. The other reading is the so-called E-type reading (e.g., Evans 1980), in which the singers who Cinswu likes must be the same as the ones that Minswu likes. Now compare (26B) with (27B) below, where the object argument site is filled with an overt pronominal.

(27)A: Minswu-ka sey myeng-uy kaswu-lul coaha-n-ta. Minswu-nom three cl-gen singer-acc like-pres-decl ‘Minswu likes three singers.’ B: Cinswu-to kutul-ul coaha-n-ta. Cinswu-also they-acc like-pres-decl ‘Cinswu likes them, too.’ [*quantiﬁcational, E-type]

The pronoun kutul ‘they’ in (27B) is and can only be anaphorically linked to the singers that Minswu likes. Thus, the sentence only yields the E-type reading. Given this, the fact

(i)A: Taro-wa sannin-no sensei-o sonkeisiteiru. Taro-top three-gen teacher-acc respects ‘Taro respects three teachers.’ B: Hanako-mo [e] sonkeisiteiru. Hanako-also respects (lit.) ‘Hanako respects, too.’ a. ‘Hanako respects three teachers, too.’ [quantiﬁcational] b. ‘Hanako also respects the same three teachers who Taro respects.’ [E-type] (ii)A: Akiu kanjian-le san-ge xuesheng. Akiu see-asp three-cl student ‘Akiu saw three students.’ B: Lisi ye kanjian-le [e]. Lisi also see-asp (lit.) ‘Lisi saw, too.’ a. ‘Lisi saw three students, too.’ [quantiﬁcational] b. ‘Lisi also saw the same three students who Akiu saw.’ [E-type]

117 that (26B) does allow the quantiﬁcational reading (as well as the E-type reading) remains mysterious if an empty pronoun, pro, indeed occupies the object argument position. On the other hand, the ellipsis analysis can easily capture the availability of the quantiﬁcational reading of the null object. Consider the two possible structural representations of (26) under the argument ellipsis analysis and the verb-stranding VP ellipsis analysis, respectively, as given in (28) and (29) below.

(28) Argument ellipsis:

VP VP

DP V DP(=[e]) V

three cl-gen...singer-acc like three cl-gen...singer-acc like

(29) Verb-stranding VP Ellipsis:

TP TP

Minswu-nom T0 Cinswu-also T0

VP T VP(=[e]) T

DP V likev DP V likev

three cl-gen...singer-acc tv three cl-gen...singer-acc tv

In (28) and (29), either of the structures for (26B) has an (indefinite) object DP containing sey myeng(-uy) ‘three cl(-gen)’, which undergoes ellipsis under identity with its overt antecedent, and can thus readily induce the quantificational reading. Similarly as in the case for the sloppy identity reading, the above ellipsis account of the quantificational reading can be further reinforced by the (VP) ellipsis-pronominal pair in English below. The sentence in (30a) involves the ellipsis of a full-fledged structure containing the numerically quantified object three singers. Thus, the quantificational reading can be naturally obtained; the sentence in (30b), on the other hand, has the pronoun them in the object position and, thus, does not allow such a quantificational reading.

(30) a. Minswu likes three singers, and Cinswu does [like three singers], too. [quantiﬁcational, E-type] b. Minswu likes three singers, and Cinswu likes them, too. [*quantiﬁcational, E-type]

118 The ellipsis analyses discussed so far, however, have been rejected by Hoji (1998, 2003) and subsequent researchers (e.g., G. Li 1998; Ahn & Cho 2010, 2011a, b; Moon 2010; Bae & Kim 2012), who have argued that the relevant null object examples can be all accounted for in terms of empty pronominals, without postulating any ellipsis mechanism.17 Consider ﬁrst the Korean null object construction in (20), repeated in (31) below.

(31)A: John 1-i caki1-uy khemphwute-lul pwuswu-ess-ta. John-nom self-gen computer-acc destroy-past-decl

‘John1 destroyed his1 computer.’ B: Peter-to [e] pwuswu-ess-ta. Peter-also destroy-past-decl (lit.) ‘Peter destroyed, too.’

a. ‘Peter destroyed his own computer, too.’ [sloppy identity]

b. ‘Peter destroyed John’s computer, too.’ [strict identity]

Hoji (1998) argues that null pronominals in East Asian languages, including Korean, can be construed as indefinite as well as definite (cf. Jaeggli 1986; Rizzi 1986). According to Hoji, the null object position [e] in (31B) is occupied by a pro, as represented in (32), that can be anaphorically linked to an indefinite argument, khemphwute ‘computer’, which corresponds to the noun head of the full DP object in the antecedent sentence in (31A). Under this ‘indefinite pro’ analysis (or ‘supplied N Head’ analysis), then, the null object sentence in (31B) is taken to be semantically equivalent to the sentence in (33B) containing the lexically overt object, khemphwute ‘computer’.18

(32) VP VP

DP V DP(=[e]) V

pro destroy caki-gen computer-acc destroy

(33)A: John 1-i caki1-uy khemphwute-lul pwuswu-ess-ta. John-nom self-gen computer-acc destroy-past-decl

‘John1 destroyed his1 computer.’

17Hoji (1998) argues against the verb-stranding VP ellipsis analysis by demonstrating that unlike English VP ellipsis, some Japanese null object constructions (i) lack a sloppy identity reading, or (ii) do not show locality effects on a sloppy reading. See Hoji (1998: 130-138) for more details of his argument. 18As pointed out by Saito (2007: 207), the postulation of an indefinite pro could be considered to be “non-standard”, since a pro is generally taken to be “a pronoun without phonetic content”, as in Chomsky (1981, 1982), and thus it should be expected to have a definite (or referential) interpretation, i.e., to be a definite pro.

119 B: Peter-to khemphwute-lul pwuswu-ess-ta. Peter-also computer-acc destroy-past-decl ‘Peter destroyed a computer, too.’

As noted by Saito (2007: 206), the interpretation of the sentence in (33B) is not exactly identical to the sloppy identity reading given in (31), ‘Pete destroyed his own computer, too’. Nevertheless, this sentence is (pragmatically) consistent with and can thus be truthfully uttered under the situation or the discourse context constituted by the above sloppy identity reading. Given this consideration, Hoji (1998) proposes that the availability of the sloppy identity reading (in his terms, the ‘sloppy-like reading’) for the null object sentence in (31B) can be analyzed in terms of pro, without recourse to an elided structure embedding the bound variable element caki-uy ‘self-gen’.19 Consider now the Korean null object construction with a quantiﬁcational reading in (26), repeated here in (34).

(34)A: Minswu-ka sey myeng-uy kaswu-lul coaha-n-ta. Minswu-nom three cl-gen singer-acc like-pres-decl ‘Minswu likes three singers.’ B: Cinswu-to [e] coaha-n-ta. Cinswu-also like-pres-decl

19Saito (2007: 207), however, argues that Hoji’s (1998) argument is signiﬁcantly undermined if the null object construction involves negation, as illustrated below.

(i)A: John 1-i caki1-uy khemphwute-lul pwuswu-ess-ta. John-nom self-gen computer-acc destroy-past-decl

‘John1 destroyed his1 computer.’ B: Peter-nun [e] an pwuswu-ess-ta. Peter-top neg destroy-past-decl (lit.) ‘Peter did not destroy.’ a. ‘Peter did not destroy his own computer.’ [sloppy identity] b. ‘Peter did not destroy John’s computer.’ [strict identity]

The negated null object sentence in (iB) can induce the sloppy identity reading, ‘Peter did not destroy his own computer’, which is compatible with the context where Peter destroyed a computer, but not his own computer. This observation, however, cannot be explained by the indeﬁnite pro analysis, since the negated sentence in (iiB) below containing the indeﬁnite object, khemphwute ‘computer’, only means that Peter did not destroy any computer at all, and thus is inappropriate under the situation/context described by the above sloppy identity reading.

(ii)A: John 1-i caki1-uy khemphwute-lul pwuswu-ess-ta. John-nom self-gen computer-acc destroy-past-decl

‘John1 destroyed his1 computer.’ B: Peter-nun khemphwute-lul an pwuswu-ess-ta. Peter-top computer-acc neg destroy-past-decl ‘Peter did not destroy a computer.’

Under the ellipsis analysis, however, the sloppy identity reading in (iB) is easily obtained by postulating an elided (VP or DP) structure containing caki ‘self’.

120 (lit.) ‘Cinswu likes, too.’

a. ‘Cinswu likes three singers, too.’ [quantiﬁcational]

b. ‘Cinswu also likes the same three singers who Minswu likes.’ [E-type]

Following the spirit of Hoji (1998), Ahn and Cho (2011a, 2011b) suggest that the null object sentence in (34B) involves an indefinite pro, as represented in (35), and thus is equivalent to the sentence in (36B) below, where the object position is filled with the indefinite nominal, kaswu ‘singer’.

(35) VP VP

DP V DP(=[e]) V

three cl-gen...singer-acc like pro like

(36)A: Minswu-ka sey myeng-uy kaswu-lul coaha-n-ta. Minswu-nom three cl-gen singer-acc like-pres-decl ‘Minswu likes three singers.’ B: Cinswu-to kaswu-lul coaha-n-ta. Cinswu-also singer-acc like-pres-decl ‘Cinswu likes a singer, too.’

According to Ahn and Cho (2011a, b), the quantiﬁcational reading in (34B) can be understood as an explicature of the sentence in (36B), which has been derived by three pragmatic adjustment processes, disambiguation, reference assignment, and enrichment (Sperber & Wilson 1986). This analysis is arguably strengthed by the fact that, since the explicature (or the quantiﬁcational) reading is pragmatically determined, it can be explicitly cancelled, as illustrated in (37).20

20According to Ahn and Cho (2011a, b), the ellipsis analysis would incorrectly predict that the relevant reading is not cancellable, as illustrated in (i), since the null object site [e] is assumed to have internal structure containing the quantiﬁer nominal sey myeng-uy kaswu ‘three singers’.

(i)A: Minswu-ka sey myeng-uy kaswu-lul coaha-n-ta. Minswu-nom three cl-gen singer-acc like-pres-decl ‘Minswu likes three singers.’ B: # Cinswu-to sey myeng-uy kaswu-lul coaha-n-ta. Haciman, ney myeng-lul Cinswu-also three cl-gen singer-acc like-pres-decl but four cl-acc coaha-n-ta. like-pres-decl ‘Cinswu likes three singers, too, but he likes four singers.’

121 (37)A: Minswu-ka sey myeng-uy kaswu-lul coaha-n-ta. Minswu-nom three cl-gen singer-acc like-pres-decl ‘Minswu likes three singers.’ B: Cinswu-to pro (=kaswu-lul) coaha-n-ta. Haciman ney myeng-ul Cinswu-also (singer-acc) like-pres-decl but four cl-acc coaha-n-ta. like-pres-decl (lit.) ‘Cinswu likes, too, but he likes four singers.’

So far, we have seen that the ellipsis and null pronominal approaches are competitive in the precise characterization of the status of null objects in Korean (and other East Asian languages). As previously mentioned in the Introduction, the primary goal of this chapter is to contribute to this debate. In the next section, I present novel empirical data from Experiment 5 that supports the ellipsis analysis for Korean.

4.3 Experiment 5

4.3.1 Research question and predictions of the experiment

In Experiment 3 in Chapter 3, with the aid of the pronoun ku ‘he’, the (un)availability of a sloppy identity reading was used as a reliable diagnostic to identify whether Korean VP anaphora kuleha ‘do so’ has internal syntactic structure, and thus to ultimately determine whether it is derived by ellipsis or it is an instance of pro-form anaphora. Experiment 5 was carried out in much the same way as in Experiment 3, in order to investigate the syntactic nature of Korean null objects. To comprehend the fundamental logic of Experiment 5, we need to recall the interpretive properties of ku found from Experiments 1 to 3. First, there exists substantial inter-speaker variation in the bindability of ku. That is, with regard to the interpretation of quantiﬁcational sentences such as (38), some Korean speakers consistently allow a bound variable reading for ku, while other Korean speakers consistently do not.

(38) Motwu-ka ku-uy cim-ul nalu-ess-ta. everyone-nom he-gen stuﬀ-acc move-past-decl ‘Everyone moved his stuﬀ.’

a. ‘Each person moved his own stuﬀ.’ b. ‘Everyone moved one particular person’s stuﬀ.’

Recall also that in sentences like (39), the pronoun ku can readily take as its antecedent a referential matrix subject (e.g., Minswu in (39)) as long as a relevant context is provided, although it strongly prefers to have a discourse antecedent.

(39) Minswu-ka ku-uy cim-ul nalu-ess-ta. Minswu-nom he-gen stuﬀ-acc move-past-decl

122 ‘Minswu moved his stuﬀ.’

a. ‘Minswu moved his own stuﬀ.’ b. ‘Minswu moved one particular person’s stuﬀ.’

Given these considerations, Experiment 5 was designed to address the following research question.

(40) research question: Does the distribution of sloppy reading for null objects follow from the distribution of the bound-variable pronoun in Korean?

To answer this research question, the availability of sloppy identity readings in null object constructions such as (41) was examined in comparison with the availability of bound variable readings in quantiﬁcational sentences such as (38).

(41) Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] Minswu-nom he-gen stuﬀ-acc move-past-conj Kiswu-also nalu-ess-ta. move-past-decl (lit.) ‘Minswu moved his stuﬀ, and Kiswu moved, too.’

a. ‘Minswu moved Minswu’s stuff, and Kiswu moved Kiswu’s stuff, too.’ [sloppy identity] b. ‘Minswu moved Minswu’s stuff, and Kiswu moved Minswu’s stuff, too.’ [strict identity]

Given that the first conjunct in (41) is intended as ‘Minswu moved Minswu’s stuff’, the ellipsis and null pronominal analyses discussed in Section 4.2 make different predictions about the correlation between the distribution of the sloppy identity readings for the null object constructions and the distribution of the quantificational binding of ku. The details of the predictions are presented below.

Predictions of the ellipsis analysis

If the Korean null objects are generated through ellipsis of a fully-ﬂedged constituent (DP or VP), which is identical to its antecedent, then the pronoun ku must be existent in the null object site [e], as represented in (42) and (43) below, and its bindability should exhibit inter-speaker variation.

123 (42) Argument ellipsis: TP TP

DP DP 1TP 1TP Minswu-nom Kiswu-also

0 0 t1 T t1 T

VP T VP T

DP V DP (=[e]) V

ku-gen stuﬀ-acc move ku-gen stuﬀ-acc move

(43) Verb-stranding VP ellipsis: TP TP

DP DP 1TP 1TP Minswu-nom Kiswu-also

0 0 t1 T t1 T

VP T VP (=[e]) T

move DP V v DP V movev

ku-gen stuﬀ-acc tv ku-gen stuﬀ-acc tv

Recall the standard view that a sloppy identity reading under ellipsis arises from a pronoun in the ellipsis site being construed as a bound variable. We would thus expect to observe variation among Korean speakers in their acceptance of sloppy identity readings in null object constructions as in (41), as well as in their acceptance of bound variable readings in quantificational sentences as in (38). Additionally, we should be able to find a high correlation between the distribution of the two readings in question. This would indicate that an individual speaker’s acceptance of the sloppy identity readings should be predictable from her acceptance of the quantificational binding of ku. For example, speakers who accept the bound variable reading in (38a) would be expected to allow the sloppy identity reading

124 in (41a), while speakers who reject the bound variable reading would be expected to disallow the sloppy identity reading.21

Predictions of the null pronominal analysis

If the Korean null objects are base-generated as indeﬁnite pros, then they would not have an internal syntactic structure, and thus the pronoun ku would never be existent in the null object site [e], as represented in (44) below.

(44) Null pronominal: TP TP

DP DP 1TP 1TP Minswu-nom Kiswu-also

0 t T0 t1 T 1

VP T VP T

DP V DP (=[e]) V

ku-gen stuﬀ-acc move pro move

Accordingly, the availability of sloppy identity readings in null object constructions as in (41) would never be influenced by the (un)bindability of ku, and thus should not correlate with the availability of bound variable readings in quantificational sentences as in (38). Recall Hoji’s (1998) claim that the null objects, as indefinite null pronominals, can license sloppy identity readings (which he calls ‘sloppy-like readings’) through pragmatic inferences. Under this view, then, the sloppy identity reading in (41a) should be expected to be unequivocally accepted by Korean speakers, once a relevant context is clearly provided in the discourse.22

21The strict identity reading in (41b) would be expected to be available to all Korean speakers, since ku in the null object site [e] would readily serve as a (co)referential pronoun, which has been standardly assumed to be the source of strict identity readings under ellipsis. 22For the same reason, we should also expect the strict identity reading in (41b) to be uniformly available to Korean speakers.

125 Figure 4.1: Screenshot of a test trial in Experiment 5

4.3.2 Methodology

4.3.2.1 Participants

Forty native Korean adult speakers living in Vancouver, Canada, who did not participate in any of the previous experiments participated in Experiment 5. They all met the same requirements for eligibility for participation as in the previous experiments.23

4.3.2.2 Task

The same truth-value judgment task was employed as in the previous experiments. That is, the participants were presented with sentences describing a context on a computer screen, followed by a target sentence. They were then asked to judge whether the target sentence truthfully described the given context by clicking 1 for ‘True’ and 0 for ‘False’ (see Figure 4.1 above).

4.3.2.3 Design and Materials

The experimental design of Experiment 5 was similar to that of Experiment 3. Each target sentence was either a null object construction such as (41) or a quantiﬁcational sentence such as (38), and each context was biased towards either bound or free (referential) interpretation of ku in the target sentence. Thus, two factors were crossed to create four experimental conditions: sentence type (NullObject vs. Quantiﬁcational) × context type (Bound

23The participants were between the ages of 20 and 27 (the mean age was 23). Most of them were university students in Korea, but came to Canada to work part-time or study English at ESL institutions temporarily. Four participants were attending colleges in Vancouver at the time of participation.

126 vs. Free). A sample set of test items are given in (45)-(48) below, where the target sentences are in boldface.24

(45) NullObject-Bound (sloppy identity reading) condition:

Minswu, Kiswu, Cinswu-ka say kiswuksa-lo isaha-ko iss-ess-ta. Minswu Kiswu Cinswu-nom new dormitory-to move-prog past-decl Minswu-ka Minswu-uy cim-ul nalu-ess-ta. Kiswu-to Kiswu-uy Minswu-nom Minswu-gen stuﬀ-acc move-past-decl Kiswu-also Kiswu-gen cim-ul nalu-ess-ta. stuﬀ-acc move-past-decl

‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuﬀ. Kiswu also moved Kiswu’s stuﬀ.’

Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] Minswu-nom he-gen stuﬀ-acc move-past-conj Kiswu-also nalu-ess-ta. move-past-decl

‘Minswu moved his stuﬀ, and Kiswu moved, too.’ (46) NullObject-Free (strict identity reading) condition:

Minswu, Kiswu, Cinswu-ka say kiswuksa-lo isaha-ko iss-ess-ta. Minswu Kiswu Cinswu-nom new dormitory-to move-prog past-decl Minswu-ka Minswu-uy cim-ul nalu-ess-ta. Kiswu-to Minswu-uy Minswu-nom Minswu-gen stuﬀ-acc move-past-decl Kiswu-also Minswu-gen cim-ul nalu-ess-ta. stuﬀ-acc move-past-decl

‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuﬀ. Kiswu also moved Minswu’s stuﬀ.’

Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] Minswu-nom he-gen stuﬀ-acc move-past-conj Kiswu-also nalu-ess-ta. move-past-decl

‘Minswu moved his stuﬀ, and Kiswu moved, too.’ (47) Quantificational-Bound condition:

24A complete set of all test items can be found in Appendix F.

127 cim-ul nalu-ess-ta. Cinswu-to Cinswu-uy cim-ul stuﬀ-acc move-past-decl Cinswu-also Cinswu-gen stuﬀ-acc nalu-ess-ta. move-past-decl

‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuff. Kiswu also moved Kiswu’s stuff. Cinswu also moved Cinswu’s stuff.’

Motwu-ka ku-uy cim-ul nalu-ess-ta. everyone-nom he-gen stuﬀ-acc move-past-decl

‘Everyone moved his stuﬀ.’ (48) Quantificational-Free condition:

Minswu, Kiswu, Cinswu-ka Thayswu-ka oki-lul kitalimye, say Minswu Kiswu Cinswu-nom Thayswu-nom coming-acc waiting new kiswuksa-lo isaha-ko iss-ess-ta. Minswu-ka Thayswu-uy cim-ul dormitory-to move-prog past-decl Minswu-nom Thayswu-gen stuff-acc nalu-ess-ta. Kiswu-to Thayswu-uy cim-ul nalu-ess-ta. move-past-decl Kiswu-also Thayswu-gen stuff-acc move-past-decl Cinswu-to Thayswu-uy cim-ul nalu-ess-ta. Cinswu-also Thayswu-gen stuff-acc move-past-decl

‘Minswu, Kiswu, and Cinswu were moving to a new dormitory, waiting for Thayswu to come. Minswu moved Thayswu’s stuff. Kiswu also moved Thayswu’s stuff. Cinswu also moved Thayswu’s stuff.’

Motwu-ka ku-uy cim-ul nalu-ess-ta. everyone-nom he-gen stuﬀ-acc move-past-decl

‘Everyone moved his stuﬀ.’

In (45), the context is consistent with the sloppy identity reading for the target null object construction, which would arise, according to the ellipsis analysis, from ku being a bound pronoun, while in (46), the context is consistent with the strict identity reading for the target null object construction, which would be attributed to ku serving as a free pronoun. In (47), the context is consistent with the bound variable reading for ku in the target quantiﬁcational sentence, while in (48), the context is consistent with the referential reading for ku in the target quantiﬁcational sentence. Based on the sample set illustrated in (45)-(48), 16 sets of test items were constructed, thus resulting in 64 test items (16 items for each of the four experimental conditions). These items were then assigned to four presentation lists according to a Latin Square design, such

128 that each list contained four items in each condition. The same 40 ﬁller items were then added to each list, some of which are given in (49)-(51) below.25

(49) Referential condition:

Minswu, Kiswu, Cinswu-ka uwntong hwu swui-ko iss-ess-ta. Minswu Kiswu Cinswu-nom exercise after rest-prog past-decl Minswu-ka Minswu-uy umlyoswu-lul masi-ess-ta. Minswu-nom Minswu-gen beverage-acc drink-past-decl

‘Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank Minswu’s beverage.’

Minswu-ka ku-uy umlyoswu-lul masi-ess-ta. Minswu-nom he-gen beverage-acc drink-past-decl

‘Minswu drank his beverage.’

(50) Nullobject-‘Object-mismatch’ condition:

Minswu, Kiswu, Cinswu-ka uwntong hwu swui-ko iss-ess-ta. Minswu Kiswu Cinswu-nom exercise after rest-prog past-decl Minswu-ka umlyoswu-lul masi-ess-ta. Kiswu-nun mwul-lul Minswu-nom beverage-acc drink-past-decl Kiswu-top water-acc masi-ess-ta. drink-past-decl

‘Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank a beverage. Kiswu drank water.’

Minswu-ka umlyoswu-lul masi-ess-ko, Kiswu-to [e] Minswu-nom beverage-acc drink-past-conj Kiswu-also masi-ess-ta. drink-past-decl

‘Minswu drank a beverage, and Kiswu drank, too.’ (constructed on the basis of J. S. Kim’s (2012) examples)

(51) Nullobject-‘Modifier-mismatch’ condition:

Minswu, Kiswu, Cinswu-ka uwntong hwu swui-ko iss-ess-ta. Minswu Kiswu Cinswu-nom exercise after rest-prog past-decl Minswu-ka chaka-wun umlyoswu-lul masi-ess-ta. Kiswu-nun ttattuthan Minswu-nom cold-adn beverage-acc drink-past-decl Kiswu-top warm umlyoswu-lul masi-ess-ta. beverage-acc drink-past-decl

25In footnote 28, I brieﬂy present and discuss the results obtained in the Nullobject-‘Modifier- mismatch’ ﬁller condition (51), which suggests that the relevant null object constructions should be analyzed as being derived by ellipsis.

129 ‘Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank a cold beverage. Kiswu drank a warm beverage.’

Minswu-ka chaka-wun umlyoswu-lul masi-ess-ko, Kiswu-to [e] Minswu-nom cold-adn beverage-acc drink-past-conj Kiswu-also masi-ess-ta. drink-past-decl

‘Minswu drank a cold beverage, and Kiswu drank, too.’ (constructed on the basis of J. S. Kim’s (2012) examples)

4.3.2.4 Procedure

The experiment was administered using PsychoPy software. The procedure was the same as in Experiments 1 to 3. Sixteen test trials (four trials per condition) and 40 ﬁller trials were presented to the participants in a uniquely generated random order. They were paid $10 each as compensation for participation.

4.3.3 Findings

Figure 4.2 summarizes mean rates of acceptance (i.e., assignment of 1 ‘True’) by condition: 33% in the Quantificational-Bound condition, 31% in the NullObject-Bound (sloppy identity reading) condition, 82% in the Quantificational-Free condition, and 79% in the NullObject- Free (strict identity reading) condition. Data were fitted to generalized linear mixed-effects models using the ‘lmer’ function of the ‘lme4’ package in the R statistical software, to analyze the participants’ responses as a function of sentence type and context type, with participants and items included as random effects. The analysis revealed a main effect of context type (coefficient estimate = 2.46, s.e. = .29, z = 8.44, p < .001): regardless of sentence type, speakers were significantly more likely to accept free reading than bound reading. However, the analysis revealed no main effect of sentence type, and no interaction between context type and sentence type: speakers were equally likely to accept bound readings for the quantificational sentences and the null object constructions; speakers were also equally likely to accept free readings for both sentence types. As in Experiments 1 to 3, all participants (n=40) were assigned into three different groups on the basis of their mean individual acceptance rates in the Quantificational-Bound condition and the NullObject-Bound (sloppy identity reading) condition: accept (> 75% acceptance: assignment of 1 to three or four target sentences), ambivalent (= 50% acceptance: assignment of 1 to two target sentences), and reject (< 25% acceptance: assignment of 1 to none or one target sentence). As a result, a bimodal distribution of the participants’ responses was observed in each Bound condition, as illustrated in Figure 4.3 below: participants tended to either always accept or always reject the quantificational

130 Quantificational NullObject 100 80 60 40 Acceptance rate(%) Acceptance 20 0 Bound Free

Figure 4.2: Mean rates of acceptance and standard errors in Experiment 5 30 Quant-Bound NullObj-Bound 25 20 15 10 Number of participants Number 5 0 Reject(0-25%) Ambivalent(50%) Accept (75-100%)

Figure 4.3: Distribution of responses in Quantiﬁcational-Bound and NullObject-Bound (sloppy identity reading) conditions in Experiment 5

131 binding interpretation for ku; participants also tended to either always accept or always reject the sloppy identity readings for the null object constructions.26 Given that inter-speaker variation was found in the Quantificational-Bound and Null Object-Bound (sloppy identity reading) conditions, a linear regression analysis was carried out to examine the correlation between the participants’ mean acceptance rates in the two Bound conditions. The analysis revealed a strong correlation (r = 0.79, t = 6.43, p <. 001), as illustrated in Figure 4.4 below. This result indicates that an individual speaker’s acceptance of the sloppy identity readings for the null object constructions is predictable from her acceptance of the bound variable readings for the quantificational sentences (and vice versa). Therefore, it is highly likely that participants who allowed the quantificational binding of ku accepted the sloppy identity readings for the null object constructions, and those who did not allow the quantificational binding of ku rejected the sloppy identity readings.27

26One might say that most of the participants did not accept the null object constructions under the sloppy identity reading contexts, as exempliﬁed by (i) below (repeated from (45)), probably because they rarely allowed the pronoun ku in the ﬁrst conjuncts to be anaphorically linked to the clause-mate subjects (e.g., Minswu in (i)).

(i) NullObject-Bound (sloppy identity reading) condition: (context: Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuﬀ. Kiswu also moved Kiswu’s stuﬀ.)

Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] nalu-ess-ta. Minswu-nom he-gen stuﬀ-acc move-past-conj Kiswu-also move-past-decl

‘Minswu moved his stuff, and Kiswu moved, too.’ Recall from footnote 28 in Chapter 2 that ku generally prefers a discourse antecedent to an intra-sentential antecedent. However, the results obtained from the filler trials for the Referential condition, as exemplified by (ii) below (repeated from (49)), revealed a high mean acceptance rate (86%), indicating that ku can readily take a clause-mate subject antecedent as long as an appropriate context is provided (see Experiment 1A in Chapter 2 for a similar result in the Referential-Overt condition).

(ii) Referential condition: (context: Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank Minswu’s beverage.)

Minswu-ka ku-uy umlyoswu-lul masi-ess-ta. Minswu-nom he-gen beverage-acc drink-past-decl

‘Minswu drank his beverage.’ Given these experimental results, it is reasonable to conclude that those participants who rejected the target null object constructions in the NullObject-Bound condition did so because they considered the first conjuncts acceptable, but the second conjuncts (i.e., the null object sentences) unacceptable. 27As predicted in Subsection 4.3.1, the mean acceptance rates in the Quantificational-Free and NullObject- Free (strict identity reading) conditions are uniformly high, thus confirming the accuracy and reliability of the results obtained in the Quantificational-Bound and NullObject-Bound (sloppy identity reading) conditions, which are the key test conditions of the experiment.

132 1.0 0.8 0.6 0.4 NullObj-Bound 0.2 0.0

0.0 0.2 0.4 0.6 0.8 1.0 Quant-Bound

Figure 4.4: Correlation between mean acceptance rates in Quantiﬁcational-Bound and NullObject-Bound (sloppy identity reading) conditions in Experiment 5

4.3.3 Discussion of Experiment 5

The ﬁndings of Experiment 5 present empirical evidence that is inconsistent with the view that the examples of null objects in Korean (and other East Asian languages) are all instances of phonologically empty pronominals. As discussed in Subsection 4.3.1, if the Korean null objects in sentences such as (41), repeated in (52) below, were indeed base-generated as (indeﬁnite) pros, atomic elements which do not have internal structure (that can accommodate the pronoun ku), then the availability of the relevant sloppy identity readings would be expected to not correlate with the availability of the bound variable construal of ku in sentences such as (38), repeated in (53) below.

(52) Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] Minswu-nom he-gen stuﬀ-acc move-past-conj Kiswu-also nalu-ess-ta. move-past-decl (lit.) ‘Minswu moved his stuﬀ, and Kiswu moved, too.’

a. ‘Minswu moved Minswu’s stuff, and Kiswu moved Kiswu’s stuff, too.’ [sloppy identity] b. ‘Minswu moved Minswu’s stuff, and Kiswu moved Minswu’s stuff, too.’ [strict identity]

133 (53) Motwu-ka ku-uy cim-ul nalu-ess-ta. everyone-nom he-gen stuﬀ-acc move-past-decl ‘Everyone moved his stuﬀ.’

a. ‘Each person moved his own stuﬀ.’ b. ‘Everyone moved one particular person’s stuﬀ.’

Contrary to the prediction of the null pronominal analysis, however, Experiment 5 have found a strong correlation between the distribution of the sloppy identity readings and the quantificational binding of ku. Furthemore, it has been demonstrated that native speakers of Korean may be sorted into two distinct groups; a group of speakers who allow both the sloppy identity readings for null objects (as in (52a)) and the quantificational binding of ku (as in (53a)), and a group of speakers who do not allow either of them. These findings can be easily accounted for by postulating that the relevant null object constructions are derived by ellipsis (i.e., PF-deletion) of an argument DP or a VP with internal structure containing ku.28,29

28As brieﬂy mentioned in footnote 25, the results obtained from the participants’ responses in the Nullobject-‘Modifier-mismatch’ ﬁller condition further support the view that some cases of Korean null objects should be analyzed as involving ellipsis, not pro. Let us consider (51) again, repeated below in (i).

(i) NullObject-‘Modifier-mismatch’ condition: (context: Minswu, Kiswu, and Cinswu were taking a rest after exercise. Minswu drank a cold beverage. Kiswu drank a warm beverage.)

Minswu-ka chakawun umlyoswu-lul masi-ess-ko, Kiswu-to [e] masi-ess-ta. Minswu-nom cold beverage-acc drink-past-conj Kiswu-also drink-past-decl

‘Minswu drank a cold beverage, and Kiswu drank, too.’

As noted by J. S. Kim (2012: 42-43), the ellipsis and null pronominal analyses yield different predictions regarding the truth-value of the target sentence in (i). Under the argument ellipsis analysis or verb-stranding VP ellipsis analysis, the null object site [e] in the second conjunct involves a full-fledged structure containing chakawun umlyoswu ‘cold beverage’, which is elided under identity with its antecedent in the first conjunct. Therefore, the whole target sentence should be judged false, since it is not compatible with the given context where Minswu drank a cold beverage, but Kiswu drank a warm beverage. According to the null pronominal analysis, on the other hand, the null object site [e] corresponds to the head noun of the full DP object in the first conjunct, umlyoswu ‘beverage’. Thus, the whole target sentence should be taken to truthfully describe the given context, since it is true that both Minswu and Kiswu drank a beverage, although the beverages consumed were of different temperatures. The results revealed that the participants rejected the target null object constructions in the ‘modifier-mismatch’ condition 93% of the time, indicating that they interpreted the null object site [e] as involving the same full DP objects as in the first conjuncts, and thus supporting the ellipsis analysis. See J. S. Kim’s (2012) TVJT experiment for similar results. 29In a similar way as discussed in footnote 29 of Chapter 2, one might maintain that the split among the Korean speakers with regard to the sloppy identity readings for the null objects could have been due to the speakers’ initial access to one of the two possible interpretations (i.e., a sloppy identity reading and a strict identity reading). Even if that were the case, the current conclusion on the syntax of Korean null objects would be still maintained. That is, it was found in Experiment 5 that an individual speaker’s acceptance of the sloppy identity readings for the null objects with antecedents containing ku was strongly correlated with her acceptance of quantifier-bound ku, and this could be attributed to each speaker’s “initial access bias” to the interpretation of an elided ku in the null objects, which would be expected to be the same as her “initial access bias” to the interpretation of the overt ku.

134 Figure 4.5: Mean rates of acceptance and standard errors in Experiment 3 and Experiment 5

At this point, one might possibly maintain that the inter-speaker variation found with the sloppy identity readings for the Korean null objects could have been due to some aspects of the task or method employed in the experiment, or due to general characteristics of Korean anaphora. However, this line of argument cannot be sustained if we review the findings of Experiment 3 in Chapter 3, which investigated the syntactic status of Korean VP anaphora kuleha ‘do so’. Consider first Figure 4.5 above, where the mean acceptances rates and standard errors in Experiment 5 can be compared to those in Experiment 3. Let us then recall that in Experiment 3, the availability of the sloppy identity reading in VP anaphora constructions such as (54) was inspected in comparison with the availability of the quantificational binding interpretation for ku, in much the same way as in Experiment 5.

(54) Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to kuleha-yess-ta. Minswu-nom he-gen stuff-acc move-past-conj Kiswu-also so.do-past-decl (lit.) ‘Minswu moved his stuff, and Kiswu did so, too.’ a. ‘Minswu moved Minswu’s stuff, and Kiswu moved Kiswu’s stuff, too.’ [sloppy identity]

135 b. ‘Minswu moved Minswu’s stuﬀ, and Kiswu moved Minswu’s stuﬀ, too.’ [strict identity]

The findings of Experiment 3 have demonstrated that while there existed considerable inter-speaker variation in the acceptance of the quantificational binding of ku, just as in Experiment 5, (nearly) all participants unequivocally accepted the sloppy identity readings for the VP anaphors. We have concluded from these findings that the Korean VP anaphors are base-generated verbal pro-forms, atomic elements that lack internal structure in the syntax. Given these considerations, if the relevant null objects were indeed instances of empty pronouns, then uniform and high acceptance of the sloppy identity readings for the null object constructions would have been expected, just as in the VP Anaphora-Bound (sloppy identity reading) condition in Experiment 3. However, this expectation has not been confirmed by the results of Experiment 5, which revealed 31% acceptance rate in the NullObject-Bound (sloppy identity reading) condition, that was derived from about one- third of the participants who consistently accepted the target null object constructions in this condition. This strongly suggests, I argue, that the null object constructions tested in Experiment 5 should be analyzed as involving ellipsis, not pro.

4.4 General discussions

To summarize, in this chapter I have provided novel empirical evidence that supports the position that some cases of null objects in Korean, a radical pro-drop language, should be attributed to the ellipsis of a constituent with full-ﬂedged structure. However, a crucial question remains unanswered as to whether the elided element in question is a DP object (Argument ellipsis) or a VP containing the DP object (Verb-stranding VP ellipsis), as illustrated in Table 4.2 below.

Source of null objects (i) Internal structure? (ii) [e] is DP or VP?

Null pronominal Absent DP Argument ellipsis Present DP Verb-stranding VP ellipsis Present VP

Table 4.2: Syntactic status of null objects

One of the well-known diagnostics to identify the ‘size’ of ellipsis is the (un)availability of the so-called ‘null adjunct reading’ (a term coined by Funakoshi 2016) for null object constructions such as (55) below, where the antecedent sentence in (55A) contains an adverb and the null object sentence in (55B) is negated.

136 (55)A: Minho-ka ppali mwul-ul masi-ess-ta. Minho-nom quickly water-acc drink-past-decl ‘Minho drank water quickly.’ B: Kiho-nun [e] masici-an-ass-ta. Kiho-top drink-neg-past-decl (lit.) ‘Kiho did not drink.’ (constructed on the basis of Oku’s (1998: 172) Japanese examples)

Crucially, the verb-stranding VP ellipsis analysis and the argument ellipsis analysis make diﬀerent predictions about the ‘recovery’ of an adjunct in the Korean null object construction above. Under the verb-stranding VP ellipsis analysis, the null object sentence in (55B) should mean that Kiho did not drink water quickly (thus implying that Kiho drank water, but he did not do so quickly). This is so since the elided element in [e] would correspond to a VP structure, which contains the manner adverb ppali ‘quickly’ as well as the DP object mwul ‘water’, as roughly represented in the following structure.30

(56) Kiho-top [VP quickly [DP water-acc] tv ](=[e]) drinkv-neg-past

According to the argument ellipsis analysis, on the other hand, such a null adjunct reading should not be available, because only the DP object mwul ‘water’, but not the adverb ppali ‘quickly’, would undergo ellipsis in [e], as illustrated in (57), and thus the whole null object sentence in (55B) should receive the ‘object only reading’ that Kiho did not drink water at all.

(57) Kiho-top quickly [DP water-acc](=[e]) drink-neg-past

To the best of my knowledge, the dominant view in the literature of East Asian languages has been that null object sentences such as (55B) are acceptable only under the interpretation in which the DP object alone is semantically ‘recovered’ in [e] (e.g., Park 1997; Oku 1998; cf. Takahashi 2008; Cheng 2011; J. S. Kim 2012), thus supporting the argument ellipsis analysis.31,32

30As discussed in Oku (1998: 173), the verb-stranding VP ellipsis analysis assumes that (‘low’) adverbs are adjoined to VP (cf. Ko 2007; Lasnik 2003). 31The null adjunct reading for (55B) was not available to any of the native Korean speakers (including myself) that I informally consulted. However, the object only reading was readily available to all of them. 32See Japanese and Chinese examples below, which have been cited by Oku (1998: 172) and Cheng (2011: 233) as evidence for the argument ellipsis analysis.

(i)A: Bill-wa kuruma-o teineini aratta. Bill-top car-acc carefully washed ‘Bill washed the car carefully.’ B: John-wa [e] arawa-nakat-ta. John-top wash-neg-past (lit.) ‘John did not wash.’ a. ‘John did not wash the car carefully.’ [*null adjunct]

137 However, the verb-stranding VP ellipsis analysis has recently been revitalized by Fu- nakoshi (2016: 5), who argues that “the null adjunct reading becomes much easier to get” if the antecedent clause and the null object clause are combined by a disjunction connective, as illustrated in the following Japanese example.

(58) Taroo-wa teineini kuruma-o arat-ta kedo, John-wa [e] araw-anak-atta. Taroo-top carefully car-acc wash-past but, John-top wash-neg-past (lit.) ‘Bill washed the car carefully, but John did not wash.’ (adapted from Funakoshi 2016: 5, ex.(16))

According to Funakoshi, the null object clause in (58) has the meaning, ‘John washed the car, but not in a careful manner’, which can be derived if the adverb teineini ‘carefully’ as well as the object kuruma ‘car’ is recovered in [e]. Funakoshi also proposes that the null adjunct reading is available “even without the disjunction connective, if a context makes the null adjunct reading appropriate”. He provides the following Japanese example to illustrate this point.

(59) Context: Taroo and Hanako washed their parents’ cars to get allowance. Taroo was thorough in his work while Hanako was not. A: Taroo-wa teineini kuruma-o arat-ta. Bill-top carefully car-acc wash-past ‘Bill washed the car carefully.’ B: Hanako-wa [e] araw-anak-atta. Hanako-ga arat-ta ato-no Hanako-top wash-neg-past Hanako-nom wash-past after-gen kuruma-wa kitanak-atta. car-top dirty-past (lit.) ‘Hanako did not wash. The car that Hanako washed was dirty.’ (Funakoshi 2016: 7, ex.(17))

Funakoshi claims that the null object sentence in (59B) can readily mean that Hanako did not wash the car carefully, with the aid of the context and the follow-up sentence “designed to favor the null adjunct reading”. Building upon his insights and arguments, an

b. ‘John did not wash the car at all.’ [object only]

(ii)A: Zhangsan henkuaide chi-wan-le fan. Zhangsan quickly eat-finish-asp rice ‘Zhangsan finished the rice quickly.’ B: Lisi ye chi-wan-le [e]. Lisi also eat-finish-asp (lit.) ‘Lisi also finished.’ a. ‘Lisi also finished the rice quickly.’ [*null adjunct] b. ‘Lisi also finished the rice.’ [object only]

138 experimental study can be constructed in order to investigate the size of ellipsis involved in Korean null object constructions, but I will leave this for future work.

139 Chapter 5

Conclusion

In this dissertation, using an experimental methodology, I have investigated the syntactic and interpretative characteristics of three Korean anaphoric devices, third-person pronouns (Chapter 2), VP anaphora (Chapter 3), and null objects (Chapter 4). In the following sections, I provide a summary of the novel empirical ﬁndings and contributions of the present study and conclude with suggestion for future work.

5.1 Summary of the major ﬁndings

In Chapter 2, the bound variable status of the third-person pronoun ku ‘he’ has been explored. I have presented three experiments designed to inspect the possibility of inter- speaker variation in the availability of bound variable construal for ku. In Experiment 1A, I tested the availability of quantificational binding of ku. It was found that some speakers of Korean consistently accepted the quantificational binding of ku, while other speakers of Korean consistently rejected it. This suggests an existence of variation across Korean speakers in the bound variable construal for ku. In Experiment 1B, I examined the availability of quantificational binding of ku in a single clause and across clauses. It was found that an individual speaker’s acceptance of quantificational binding in a single clause is highly predictable from her acceptance of quantificational binding across a clause boundary. This suggests that clause type does not affect the bound variable construal for ku. In Experiment 2, the availability of quantificational binding of ku was examined in two different test sessions, which took place one month apart. I found a strong correlation between a speaker’s acceptance of quantificational binding of ku in the two sessions. This suggests that Korean speakers exhibit the same judgment over time on the bound variable construal of ku. Based on all these findings, I have come to the conclusion that there is indeed variation across Korean speakers regarding the availability of bound variable construal for ku.

140 I have also attempted to provide a principled account for why and how the phenomenon of inter-speaker variation arises and exists in Korean and how it can be captured in binding- theoretic terms. Taking into consideration the historical background of ku and its present status, I have concluded that child learners of Korean may not receive sufficient evidence regarding ku from the primary language input data. Given this, adopting Han et al.’s (2007) two-grammar hypothesis and Déchaine and Wiltschko’s (2002) pronominal typology, I have proposed that some speakers of Korean randomly acquire FP ku, which complies with the “pronominal grammar”, while other speakers of Korean acquire DP ku, which complies with the “demonstrative grammar”. Chapter 3 has investigated the syntactic nature of VP anaphora in Korean. I have presented two experiments designed to inspect the presence/absence of internal structure within Korean VP anaphora. In Experiment 3, building upon the finding that there is inter- speaker variation in the quantificational binding of ku (i.e., the bound variable construal for ku), I examined the availability of sloppy readings for VP anaphors with antecedents containing ku. Given the standard view that the sloppy reading in ellipsis is due to a pronoun in the ellipsis site being bound, if VP anaphors have elided structure that hosts ku, the distribution of sloppy readings for them should correlate with that of quantificational binding of ku. Such a correlation, however, is not expected if they are pro-forms, which do not host elided material (and thus not ku). This correlation was not found in Experiment 3. In Experiment 4, I examined the acceptability of extraction out of VP anaphora in the short distance and long distance scrambling contexts. If VP anaphors are instances of ellipsis, then object scrambling should be possible in both the short distance and long distance contexts, since they contain internal syntactic structure that can host a trace of movement. If VP anaphors are base-generated pro-forms, then object scrambling should not be possible in either the short distance or long distance contexts, since they do not have internal structure to host a trace of movement. It was found that sentences involving either short or long distance scrambling out of VP anaphora are significantly likely to be judged less acceptable than their corresponding non-scrambled sentences, thus suggesting that extraction out of VP anaphora is hardly available in Korean. Based on all these findings, I have claimed that Korean VP anaphors are verbal pro-forms. Chapter 4 has investigated the syntactic nature of null objects in Korean. I presented an experiment designed to inspect the presence/absence of internal structure within null objects in Korean. In a similar way as in Experiment 3, in Experiment 5, I examined the availability of sloppy readings for null objects with antecedents containing ku. If null objects have elided structure that hosts ku, an individual speaker’s acceptance of sloppy readings for null objects is highly predictable from her acceptance of quantificational binding of ku. Such a correlation is not expected if they are null pronouns that do not host elided material (and thus not ku). This correlation was found in Experiment 5. If null objects were null pro-forms, high acceptance of sloppy readings for them would have been expected, just as

141 in the VP Anaphora-Bound (sloppy identity reading) condition in Experiment 3, where VP anaphors have been found to be overt pro-forms. However, this expectation was not conﬁrmed by the results of Experiment 5. Based on all these ﬁndings, I have claimed that null objects in Korean are instances of ellipsis.

5.2 Contributions

The present study has provided novel empirical evidence that there exists inter-speaker variation regarding the availability of bound variable construal for Korean third-person pronoun ku, as a synchronically active linguistic phenomenon. As mentioned in Chapter 2, in previous studies on bound variable anaphora in Korean, much less attention has been paid to ku than the long-distance anaphor caki ‘self’. Moreover, the lack of scholarly consensus regarding the bound variable status of ku has been highly indicative of a possibility of the existence of the inter-speaker variation phenomenon, but no attempt has been made to verify this possibility experimentally. Therefore, one signiﬁcant contribution of the present study is that it has established a solid and extensive empirical base for obtaining a more precise and complete picture of the interpretative status of Korean (third-person) pronouns. Another crucial contribution of the present study is that it has demonstrated that the inter-speaker variation in the availability of bound variable construal for ku can play a crucial role in investigating the syntax of Korean anaphora. As mentioned in Chapter 3, in previous studies on anaphora, the availability of sloppy readings has been taken as a problematic diagnostics for the presence/absence of internal structure of an anaphoric item (e.g., Merchant 2013a, 2013b). However, the present study has shown that the same does not hold true for Korean anaphora. In Experiment 3 and Experiment 5, based on the assumption that ku in the elided structure should pattern with the overt ku in terms of bindablity, the availability of sloppy readings has been used as a reliable tool to determine the syntactic identity of Korean VP anaphors and null objects. In light of this, I hope that the present study will serve as a useful starting point for any future research on the syntax of other Korean anaphoric items.

5.3 Suggestion for future work

According to Déchaine and Wiltschko’s (2002) theory, DPs (including pro-DPs) are demonstrably deﬁnite and have the binding-theoretic status of R-expressions, so they are never expected to have bound variable interpretations, because of Principle C of the binding theory. Indeed, English deﬁnite descriptions, as full DPs, cannot generally be bound, as illustrated in (1).

(1) a.* Every boy 1 thinks the/that/this boy1 is smart.

142 b.* Every boy 1 loves the/that/this boy1’s girlfriend.

However, sentences such as (2) have been used in the literature to argue that English deﬁnite descriptions can be construed as bound variables.1

(2) a. Every logician1 was walking with a boy near that logician1’s house. (Evans 1977: 491)

b. John criticized every senator1 in private while praising the bastard1 in public. (Hornstein & Weinberg 1988: 149)

Taking the hypothesis that binding applies only to functional items, Noguchi (1997) argues that D-pronouns (e.g., English he) can be construed as bound variables, which are claimed to occupy the functional node D. He also claims that the definite articles and demonstratives, as functional items, can be bound and thus the definite descriptions headed by them can be bound, too. However, he points out that not all definite descriptions can be bound, as illustrated by the contrast in (3), since they are sensitive to certain factors or conditions regarding bindability that are yet to be discovered.

(3) a. Every boy1 dates a girl who adores that boy1.

b.* Every boy 1 dates a girl who adores this boy1. (Noguchi 1997: 785)

The crucial point Noguchi makes is that though not a sufficient condition, being a functional item is a necessary condition for an item to be bound. This is in stark contrast with D&W’s theory in which the DP projection itself is not expected to be bound at all. Moreover, he argues that it is no surprise that the Korean personal pronoun ku is claimed to be capable of functioning as a bound variable, since it is a member of the paradigm of Korean demonstrative determiners. This obviously contradicts the proposed pro-DP analysis of ku, in which the DP ku complies with the “demonstrative grammar” and thus cannot be construed as a bound variable. Given the conflict regarding the availability of bound variable construal for definite descriptions (or DPs), I propose to conduct an experiment to investigate how Korean speakers judge the definite descriptions headed by the demonstrative determiner ku in the bound variable reading contexts. Unfortunately, there are no studies, to my knowledge, in the literature that provide Korean data regarding the issue in question. At this point, it is worth noting Hoji’s (1990) claim that the Japanese demonstrative so corresponds to the Korean demonstrative ku, as illustrated in the following comparison between the demonstrative paradigms in Japanese and Korean (-no is a genitive marker in Japanese).

(4) Japanese Korean a. kono hon i chayk ‘this book’ (close to both speaker and hearer)

1Hornstein and Weinberg (1988) explain that in (2b), the bastard is bound by every senator at LF and is construed as a bound variable.

143 b. sono hon ku chayk ‘that book’ (far from speaker and close to hearer) c. ano hon ce chayk ‘that book’ (far from both speaker and hearer)

Additionally, Nishigauchi (1990), Hoji (1990), and Noguchi (1997) argue that the Japanese sono+N can be construed as a bound variable, as illustrated in (5).

(5) a. [Dono daigaku]1-mo [sono daigaku]1-no gengogakusya-o tairyooni every university-also that university-gen linguist-acc many kubinisita. ﬁred

‘Every university1 ﬁred that university’s1 linguists by a large number.’ (Hoji 1990: 10)

b. [Dono kaisya]1-mo [sono kaisya]1-ga itiban-da to omotte-iru. every company-also that company-nom best-cop comp think-pres

‘Every company1 thinks that that company1 is the best.’ (Noguchi 1997: 786)

Thus, the test sentences for the experiment can be adapted from the Japanese sentences with sono+N with bound variable interpretations, as exempliﬁed in (6).

(6) a. [Motun tayhakkyo]-ka [ku tayhakkyo]-uy enehakca-lul taylyangulo every university-nom that university-gen linguist-acc many haykoha-yess-ta. ﬁre-past-decl ‘Every university ﬁred that university’s linguists by a large number.’ b. [Motun hoysa]-ka [ku hoysa]-ka choyko-lako sayngkakha-n-ta. every company-nom that company-nom best-comp think-pres-decl ‘Every company thinks that that company is the best.’

The experiment will employ a similar TVJ task as in the earlier experiments and include the test trials for the quantificational-overt condition used in Experiment 1. Thus, the experiment will be able to examine (i) whether there exists inter-speaker variation regarding the bound variable construal for the pronoun ku and, if so, (ii) how the two groups of speakers judge the bound variable construal for the definite descriptions headed by the demonstrative ku. Note that unlike personal pronoun ku, child learners of Korean are provided with sufficient input regarding demonstrative ku, and thus it can be hypothesized that they may acquire a uniform demonstrative ku. Assuming that the demonstrative ku is the head of DP (Kang 2001; Chang 2009, among others), the current proposed analysis based on Déchaine and Wiltschko’s (2002) theory predicts that all Korean speakers should disallow the definite descriptions to have a bound variable interpretation, regardless of whether they allow a bound variable construal of the pronoun ku (or whether they have pro-DP ku or pro-FP ku). If this prediction is not borne out, the proposed theoretical account on the variable status of the Korean third-person pronoun will have to be reconsidered.

144 Bibliography

Abney, Steven. 1987. The English noun phrase in its sentential aspect. Doctoral Disserta- tion, MIT, Cambridge, Massachusettes.

Aelbrecht, Lobke. 2010. The syntactic licensing of ellipsis. Amsterdam: John Benjamins Publishing.

Ahn, Hee-Don, and Sungeun Cho. 2006. Layered nominal structures: Implications for caseless nominals. Korean Journal of Linguistics 31:165–185.

Ahn, Hee-Don, and Sungeun Cho. 2010. More on the absence of CP ellipsis. Studies in Generative Grammar 20:137–148.

Ahn, Hee-Don, and Sungeun Cho. 2011a. On sloppy-like interpretation of null arguments. Linguistic Research 28:471–492.

Ahn, Hee-Don, and Sungeun Cho. 2011b. Notes on apparent DP ellipsis. Korean Journal of Linguistics 36:457–471.

An, Sojin. 2008. The function of third-person pronouns ‘ku, kunye’ [in Korean]. Korean Linguistics 38:145–164.

Aoun, Joseph, and Norbert Hornstein. 1992. Bound and referential pronouns. In Logi- cal structure and linguistic structure, ed. C.-T. James Huang and Robert May, 1–23. Dordrecht: Kluwer.

Bae, Eun-Gyeong, and Jeong-Suk Kim. 2012. On missing objects: Model-interpretive anaphora or ellipsis. Studies in Modern Grammar 70:49–71.

Bak, Jae-Hee. 2015. An research on the 3rd person pronoun ‘ku’ [in Korean]. Hanminjoke- munhak 71:103–128.

Baltin, Mark, Rose-Marie Déchaine, and Martina Wiltschko. 2013. The structural hetero- geneity of pronouns. Paper presented at the Linguistic Society of America 2013 Annual Meeting, Boston.

Barski, Ewelina. 2013. Spanish and Polish heritage speakers in Canada: The overt pronoun constraint. Doctoral Dissertation, University of Western Ontario, Canada.

Bates, Douglas, Martin Maechler, and Bolker Bolker. 2012. lme4: Linear mixed-eﬀects models using S4 classes. R Package Version 0.999999-0. (http://CRAN.R-project. org/package=lme4).

145 Bentzen, Kristine, Jason Merchant, and Peter Svenonius. 2013. Deep properties of surface pronouns: pronominal predicate anaphors in Norwegian and German. The Journal of Comparative Germanic Linguistics 16:97–125.

Bosch, Peter, Tom Rozario, and Yufan Zhao. 2003. Demonstrative pronouns and personal pronouns. German der vs. er. In Proceedings of the EACL 2003 Workshop on the Computational Treatment of Anaphora. Budapest, Hungary: EACL.

Bruening, Benjamin. 2018. Passive do so. Natural Language and Linguistic Theory 36:1–49.

Büring, Daniel. 2005. Binding theory. Cambridge, UK: Cambridge University Press.

Busquets, Joan. 2018. Fer-ho anaphora in Catalan: Semantic and discourse properties. Journal of Psycholinguistic Research 47:307–324.

Carminati, Maria N. 2013. Bound variable reading in Italian: pro, the over pronoun and PRO. A psycholinguistic study. Paper presented at the Workshop on Impact of Pronom- inal Form on Interpretation, University of Tübingen, Germany.

Cecchetto, Carlo, and Orin Percus. 2006. When we do that and when we don’t: A contrastive analysis of VP ellipsis and VP anaphora. In Phases of interpretation, ed. Mara Frascarelli, 71–105. Berlin: Mouton de Gruyter.

Chang, Soo-Jung. 2009. Nominal structure and interpretation. Doctoral Dissertation, Uni- versity of Georgia, Athens, Georgia.

Cheng, Hsu-Te. 2011. Argument ellipsis in Mandarin Chinese. In Proceedings of the 23rd North American Conference on Chinese Linguistics, ed. Zhuo Jing-Schmidt, 224–240. University of Oregon, Eugene.

Cheng, Hsu-Te. 2013. Argument ellipsis, classiﬁer phrases, and the DP parameter. Doctoral Dissertation, University of Connecticut, Storrs.

Cho, Dong-In. 1996. Anaphor or pronominal? Language Research 32:621–636.

Cho, Sook-Whan. 1992. The syntax and acquisition of ‘kyay’(‘s/he’) and ‘caki’ (‘self’). Studies in Generative Grammar 2:361–392.

Cho, Young-Mee Yu, and Ki-Sun Hong. 1988. Evidence for the VP constituent from child Korean. In Papers and reports on child language development 27 , 31–38. Stanford, California: Department of Linguistics, Stanford University.

Choe, Hyon-Sook. 1988. Restructuring parameters and complex predicates: A transformational approach. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Choi, Ki-yong. 2013. A look on 3rd person referring expression ku [in Korean]. Studies in Generative Grammar 23:527–558.

Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht, the Netherlands: Foris.

Chomsky, Noam. 1982. Some concepts and consequences of the theory of government and binding. Cambridge, Massachusettes: MIT press.

146 Chomsky, Noam. 1984. Knowledge of language: Its nature, origin, and use. Cambridge, Massachusettes: MIT press.

Chomsky, Noam. 1986. Barriers. Cambridge, Massachusettes: MIT press.

Chomsky, Noam, and Howard Lasnik. 1993. Principles and parameters theory. In Syn- tax: An international handbook of contemporary research, ed. Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld, and Theo Vennemann, 505–569. Berlin: Mouton de Gruyter. Reprinted in Noam Chomsky, The Minimalist Program, Cambridge, Mas- sachusettes: MIT Press, 1995.

Chung, Chan. 1999. A nonconfigurational approach to the weak crossover effect in Ko- rean. In Proceedings of the 13th Pacific Asia Conference on Language, Information and Computation, ed. Jhing-Fa ang and Chung-Hsien Wu, 131–142. Tainan, Taiwan: National Cheng Kung University.

Chung, Sandra, William Ladusaw, and James McCloskey. 2011. Sluicing (:) between structure and inference. In Representing language: Essays in honor of Judith Aissen, ed. Rodrigo Gútierrez-Bravo, Line Mikkelsen, and Eric Potsdam, 31–50. Santa Cruz, Cal- ifornia: Linguistics Research Center, University of California, Santa Cruz.

Cole, Peter. 1987. Null objects in universal grammar. Linguistic Inquiry 18:597–612.

Cornish, Francis. 1986/2005. Anaphoric relations in English and French: A discourse perspective. London, UK: Croom Helm.

Costa, João, and Ana Maria Martins. 2011. On focus movement in European Portuguese. Probus 23:217–245. van Craenenbroeck, Jeroen, and Jason Merchant. 2013. Elliptical phenomena. In Cam- bridge handbook of generative syntax, ed. Marcel Den Dikken, 701–745. Cambridge, UK: Cambridge University Press.

Crain, Stephen, and Cecile McKee. 1985. The acquisition of structural restrictions on anaphora. In Proceedings of the 15th Annual Meeting of North East Linguistic Soci- ety, ed. Stephen Berman, Jae-Woong Choe, and Joyce McDonough, 94–110. Amherst, Massachusettes: GLSA, University of Massachusettes, Amherst.

Crain, Stephen, and Rosalind Thornton. 1998. Investigations into Universal Grammar: A guide to experiments on the acquisition of syntax and semantics. Cambridge, Mas- sachusettes: MIT Press.

Culicover, Peter W, and Ray Jackendoﬀ. 2005. Simpler syntax. Oxford, UK: Oxford University Press.

Dalrymple, Mary. 2005. Against reconstruction in ellipsis. In Ellipsis and nonsentential speech, ed. Reinaldo Elugardo and Robert J. Stainton, 31–55. Dordrecht, the Nether- lands: Springer.

Déchaine, Rose-Marie, and Martina Wiltschko. 2002. Decomposing pronouns. Linguistic Inquiry 33:409–442.

147 Déchaine, Rose-Marie, and Martina Wiltschko. 2003. On pro-nouns and other “pronouns”. In From NP to DP, vol. 1: The syntax and semantics of noun phrases, ed. Martine Coene and Yves D’hulst, 71–90. Amsterdam: John Benjamins Publishing.

Depiante, Marcela Andrea. 2000. The syntax of deep and surface anaphora: A study of null complement anaphora and stripping/bare argument ellipsis. Doctoral Dissertation, University of Connecticut, Storrs.

Dixon, Robert. 1977. A grammar of Yidiny. Cambridge, UK: Cambridge University Press.

Doron, Edit. 1999. V-movement and VP ellipsis. In Fragments: Studies in ellipsis and gapping, ed. Shalom Lappin and Elabbas Benmamoun, 124–140. Oxford, UK: Oxford University Press.

Dubinsky, Stanley, and Robert Hamilton. 1998. Epithets as antilogophoric pronouns. Lin- guistic Inquiry 29:685–693.

Elbourne, Paul. 2005. Situations and individuals. Cambridge, Massachusettes: MIT Press.

Elbourne, Paul. 2008. Demonstratives as individual concepts. Linguistics and Philosophy 31:409–466.

Evans, Gareth. 1980. Pronouns. Linguistic Inquiry 11:337 – 362.

Farrell, Patrick. 1990. Null objects in Brazilian Portuguese. Natural Language and Linguistic Theory 8:325–346.

Fiengo, Robert. 1974. Semantic conditions on surface structure. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Fonseca-Greber, Bonnie. 2002. Swiss colloquial french and the overt pronoun constraint. In Arizona working papers in second language acquisition and teaching, ed. Claudia Kost and Senta Gortler, 104–113. Tuscon, Arizona: Department of Linguistics, University of Arizona.

Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, Massachusettes: MIT Press.

Frazier, Lyn. 2013. A recycling approach to processing ellipsis. In Diagnosis in syntax, ed. Lisa Lai-Shen Cheng and Norbert Corver, 485–501. Oxford: Oxford University Press.

Fu, Jungqi, Thomas Roeper, and Hagit Borer. 2001. The VP within process nominals: Evidence from adverbs and the VP anaphor do-so. Natural Language and Linguistic Theory 19:549–582.

Fujiwara, Yoshiki. 2017. V-stranding VP-ellipsis in child Japanese. In Proceedings of the 41st Annual Boston University Conference on Language Development, ed. Maria LaMendola and Jennifer Scott, 263–272. Somerville, Massachusettes: Cascadilla Press.

Fukaya, Teruhiko, and Hajime Hoji. 1999. Stripping and sluicing in Japanese and some implications. In Proceedings of the 18th West Coast Conference on Formal Linguistics, ed. Danny, Fox, Jon Nissenbaum, Sonya Bird, Andrew Carnie, Jason D. Haugen, and Peter Norques, 145–158. Somerville, Massachusettes: Cascadilla Press.

148 Fukui, Naoki. 1988. Deriving the diﬀerences between English and Japanese: A case study in parametric syntax. English Linguistics 5:249–270.

Funakoshi, Kenshi. 2014. Syntactic head movement and its consequences. Doctoral Disser- tation, University of Maryland, College Park.

Funakoshi, Kenshi. 2016. Verb-stranding verb phrase ellipsis in Japanese. Journal of East Asian Linguistics 25:113–142.

Garnham, Alan, and Jane Oakhill. 1996. The mental models theory of language comprehension. In Models of understanding text, ed. Bruce K. Britton and Arthur C. Graesser, 313–339. Hillsdale, New Jersey: Erlbaum.

Gazdar, Gerald. 1982. Phrase structure grammar. In The nature of syntactic representation, ed. Gerald Gazdar, Pauline Jacobson, and Geoﬀrey Pullum, 131–186. Dordrecht, the Netherlands: Reidel Publishing Company.

Goldberg, Lotus Madelyn. 2005. Verb-stranding VP ellipsis: A cross-linguistic study. Doc- toral Dissertation, McGill University.

Gordon, Peter. 1998. The truth-value judgment task. Cambridge, Massachusettes: MIT Press.

Gribanova, Vera. 2013. Verb-stranding verb phrase ellipsis and the structure of the Russian verbal complex. Natural Language and Linguistic Theory 31:91–136.

Grimshaw, Jane. 1979. Complement selection and the lexicon. Linguistic Inquiry 10:279– 326.

Grinder, John, and Paul Postal. 1971. Missing antecedents. Linguistic Inquiry 2:269–312.

Grodzinsky, Yosef, and Tanya Reinhart. 1993. The innateness of binding and coreference. Linguistic Inquiry 24:69–101.

Gürel, Ayse. 2006. L2 acquisition of pragmatic and syntactic constraints in the use of overt and null subject pronouns. In Inquiries in linguistic development, ed. Roumyana Slabakova, Silvina Montrul, and Philippe Prevost, 259–282. Amsterdam: John Ben- jamin Publishing.

Ha, Seungwan. 2010. Re-binding puzzles in Korean VP-anaphora. Korean Journal of Linguistics 35:471–487.

Haddican, Bill. 2007. The structural deﬁciency of verbal pro-forms. Linguistic Inquiry 38:539–547.

Haegeman, Liliane. 1990. Non-overt subjects in diary contexts. In Grammar in progress, ed. Joan Mascaro and Marina Nespor, 167–174. Berlin: Mouton de Gruyter.

Haegeman, Liliane, and Tabea Ihsane. 1999. Subject ellipsis in embedded clauses in English. English Language and Linguistics 3:117–145.

Haïk, Isabelle. 1987. Bound VPs that need to be. Linguistics and Philosophy 10:503–530.

149 Hallman, Peter. 2004. Constituency and agency in vp. In Proceedings of the 23rd West Coast Conference on Formal Linguistics, ed. Vineeta Chand, Ann Kelleher, Angelo J. Rodríguez, and Benjamin Schmeiser, 304–317. Somerville, Massachusettes: Cascadilla Press.

Hallman, Peter. 2013. Predication and movement in passive. Lingua 125:76–94.

Han, Chung-hye, Jeﬀrey Lidz, and Julien Musolino. 2007. V-raising and grammar competition in Korean: Evidence from negation and quantiﬁer scope. Linguistic Inquiry 38:1–47.

Han, Chung-hye, Julien Musolino, and Jeﬀrey Lidz. 2016. Endogenous sources of variation in language acquisition. Proceedings of the National Academy of Sciences 113:942–947.

Han, Chung-hye, and Dennis Ryan Storoshenko. 2012. Semantic binding of long-distance anaphor caki in Korean. Language 88:764–790.

Han, Na-Rae. 2006. Korean zero pronouns: analysis and resolution. Doctoral Dissertation, University of Pennsylvania.

Hankamer, Jorge, and Ivan Sag. 1976. Deep and surface anaphora. Linguistic Inquiry 7:391–428.

Hara, Takaaki. 2002. Bound variable interpretation and the degree of accessibility. In Student Organization of Linguistics in Europe (ConSOLE) IX, ed. Marjo van Koppen, Erica Thrift, Erik Jan van der Torre, and Malte Zimmermann, 82–95. Leiden, the Netherlands: Student Organisation of Linguistics in Europe, Leiden University.

Hardt, Daniel. 1993. Verb phrase ellipsis: Form, meaning, and processing. Doctoral Disser- tation, University of Pennsylvania.

Hasegawa, Nobuko. 1985. On the so-called “zero pronouns” in Japanese. The Linguistic Review 4:289–342.

Heim, Irene, and Angelika Kratzer. 1998. Semantics in generative grammar. Oxford, UK: Blackwell Publishing.

Hoji, Hajime. 1985. Logical form constraints and conﬁgurational structures in Japanese. Doctoral Dissertation, University of Washington.

Hoji, Hajime. 1990. On the so-called overt pronouns in Japanese and Korean. In 7th In- ternational Conference on Korean Linguistics, ed. Eung-Jin Baek, 61–78. International Circle of Korean Linguistics and Osaka University of Economics and Law.

Hoji, Hajime. 1998. Null object and sloppy identity in Japanese. Linguistic Inquiry 29:127– 152.

Hoji, Hajime. 2003. Surface and deep anaphora, sloppy identity, and experiments in syntax. In Anaphora: A reference guide, ed. Andrew Barss, 172–236. Oxford, UK: Blackwell Publishing.

150 Hoji, Hajime, Satoshi Kinsui, Yukinori Takubo, and Ayumi Ueyama. 1999. Demonstratives, bound variables, and reconstruction eﬀects. In Proceedings of the Nanzan GLOW, the 2nd GLOW Meeting in Asia, ed. Mamoru Saito, Yasuaki Abe, Hiroshi Aoyagi, Masatake Arimoto, Keiko Murasugi, and Tatsuya Suzuki, 141–158. Nagoya, Japan: Department of Linguistics, Nanzan University.

Hong, Sungshim. 1985. A-binding and A’-binding in Korean. Doctoral Dissertation, Uni- versity of Connecticut, Storrs.

Houser, Michael John. 2010. The syntax and semantics of do so anaphora. Doctoral Dissertation, University of California, Berkeley.

Houser, Michael John, Line Mikkelsen, and Maziar Toosarvandani. 2007. Verb phrase pronominalization in Danish: Deep or Surface Anaphora? In Proceedings of the 34th Western Conference on Linguistics, ed. Erin Bainbridge and Brian Agbayani, 183–195. Fresno, California: Department of Linguistics, California State University, Fresno.

Huang, C.-T. James. 1982. Logical relations in Chinese and the theory of grammar. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Huang, C.-T. James. 1984a. On the distribution and reference of empty pronouns. Linguistic Inquiry 531–574.

Huang, C.-T. James. 1984b. On the typology of zero anaphora. Language Research 20:85– 106.

Huang, C.-T. James. 1987. Remarks on empty categories in Chinese. Linguistic Inquiry 321–337.

Huang, C.-T. James. 1988. Comments on Hasegawa’s paper. In Proceedings of Japanese syntax workshop: Issues on empty categories, ed. Wako Tawa and Minerahu Nakayama, 77–93. New London, Connecticut: Japanese Program, Connecticut College.

Huang, C.-T. James. 1991. Remarks on the status of the null object. In Principles and parameters in comparative grammar, ed. Robert Freiden, 56–76. Cambridge, Mas- sachusettes: MIT press.

Huang, Yan. 1994. The syntax and pragmatics of anaphora: A study with special reference to Chinese. Cambridge, UK: Cambridge University Press.

Hwang, Shin-Ja. 1991. Terms of address in Korean and American cultures. Intercultural Communication Studies 1:117–136.

Im, Hong-Bin. 1987. Study on Korean reﬂexive [in Korean]. Seoul, Korea: Shingu Mwun- hwasa.

Im, Hong-Bin. 1998. Problems of DP hypothesis [in Korean]. In The in-depth Korean grammar: The grammar of noun phrase and postpositional phrase, ed. Hong-Bin Im. Seoul, Korea: Taehaksa.

Ionin, Tania, Soondo Baek, Eunah Kim, Heejeong Ko, and Kenneth Wexler. 2012. That’s not so diﬀerent from the: Deﬁnite and demonstrative descriptions in second language acquisition. Second Language Research 28:69–101.

151 Jackson, Kyuseek Hwang. 2008. The eﬀect of information structure on Korean scrambling. Doctoral Dissertation, University of Hawai‘i, M¯anoa.

Jaeggli, Osvaldo. 1986. Arbitrary plural pronominals. Natural Language and Linguistic Theory 4:43–76.

Jaeggli, Osvaldo, and Kenneth Saﬁr. 1989. The null subject parameter and parametric theory. In The null subject parameter, ed. Osvaldo Jaeggli and Kenneth Saﬁr, 1–44. Dordrecht, the Netherlands: Springer.

Johnson, Kyle. 2001. What VP ellipsis can do, and what it can’t, but not why. In The handbook of contemporary syntactic theory, ed. Mark Baltin and Chris Collins, 439–479. Malden, Massachusettes: Blackwell Publishing.

Kang, Beom-Mo. 1988a. Unbounded reﬂexives. Linguistics and Philosophy 11:415–456.

Kang, Jennifer Yusun. 2004. Telling a coherent story in a foreign language: Analysis of Korean EFL learners’ referential strategies in oral narrative discourse. Journal of Pragmatics 36:1975–1990.

Kang, Myung-Yoon. 1988b. Topics in Korean syntax: Phrase structure, variable binding and movement. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Kang, Myung-Yoon. 2001a. DP hypothesis and Korean nominal phrases [in Korean]. Han- kwukehak 13:15–51.

Kang, Nam-Kil. 2001b. Pronouns and competition. Studies in Modern Grammar 25:91–107.

Kang, Sahie. 1990. Discourse conditions and Korean anaphora. Doctoral Dissertation, University of Florida, Gainesville.

Kanno, Kazue. 1997. The acquisition of null and overt pronominals in Japanese by English speakers. Second Language Research 13:265–287.

Kasai, Hironobu. 2014. On the nature of null clausal complements in Japanese. Syntax 17:168–188.

Katada, Fusa. 1991. The LF representation of anaphors. Linguistic Inquiry 22:287–313.

Kehler, Andrew, and Gregory Ward. 1999. On the semantics and pragmatics of identiﬁer so. In The semantics/pragmatics interface from diﬀerent points of view (Current research in the semantics/pragmatics interface series, Volume I), ed. Ken Turner, 233–256. Amsterdam: Elsevier.

Kehler, Andrew, and Gregory Ward. 2004. Constraints on ellipsis and event reference. In Handbook of pragmatics, ed. Laurence R. Horn and Gregory Ward, 383–403. Oxford, UK: Blackwell Publishing.

Keller, Frank, Subahshini Gunasekharan, Neil Mayo, and Martin Corley. 2009. Timing accuracy of web experiments: A case study using the WebExp software package. Behavior Research Methods 41:1–12.

152 Kennedy, Christopher. 2003. Ellipsis and syntactic representation. In The interfaces: De- riving and interpreting omitted structures, ed. Kerstin Schwabe and Susanne Winkler, 29–53. Amsterdam: John Benjamin Publishing.

Kim, Chi-hong. 1984. Dong-In Kim’s essays in criticism [in Korean]. Seoul, Korea: Samy- oungsa.

Kim, Dong-In. 1948. 30 years of my writing life [in Korean]. Seoul, Korea: Sinchenci.

Kim, Haeyeon. 1989. Nominal reference in discourse: Introducing and tracking referents in Korean spoken narratives. Harvard Studies in Korean Linguistics 3:431–444.

Kim, Hana. 2007. The interpretation of English pronouns by native speakers of Korean. Master’s Thesis, Southern Illinois University, Carbondale.

Kim, Hyung-chul. 1981. Third-person pronoun [in Korean]. Literature and Language 2:3–14.

Kim, Jeong-Seok. 1997a. Syntactic focus movement and ellipsis: A minimalist approach. Doctoral Dissertation, University of Connecticut, Storrs.

Kim, Ji-Hye, and James Yoon. 2009. Long-distance bound local anaphors in Korean? An empirical study of the Korean anaphor caki-casin. Lingua 119:733–755.

Kim, Jin-Sook. 2012. Comprehension of elided phrases in Korean and English: VP-ellipsis, null object constructions, and one-substitution. Doctoral Dissertation, University of Hawai‘i, M¯anoa.

Kim, Jong-Bok. 2016a. The syntactic structures of Korean. Cambridge, UK: Cambridge University Press.

Kim, Jong-Bok, and Peter Sells. 2008. English syntax: An introduction. Stanford, California: CSLI Publications.

Kim, Joung-Ran. 1992. Korean pronouns caku and ku: The grammatical distinction between selective and unselective binding. Master’s Thesis, University of Maryland, College Park.

Kim, Jungho, Masatoshi Koizumi, Naho Ikuta, Yuichiro Fukumitsu, Naoki Kimura, Kazuki Iwata, Jobu Watanabe, Satoru Yokoyama, Shigeru Sato, Kaoru Horie, and Ryuta Kawashima. 2009. Scrambling eﬀects on the processing of Japanese sentences: An fMRI study. Journal of Neurolinguistics 22:151–166.

Kim, Ock-Hwan, and Yoshihisa Kitagawa. 2010. Selective reproduction in NP-ellipsis. In Proceedings of Japanese/Korean Linguistics 19 , ed. Cheon Sang-Yee, 125–140. Stan- ford, California: CSLI Publications.

Kim, Soowon. 1999. Sloppy/strict identity, empty objects, and NP ellipsis. Journal of East Asian Linguistics 8:255–284.

Kim, Sun-Hee. 2010. Null arguments and the theory of pro. Korean Journal of Linguistics 35:563–590.

153 Kim, Yong Ha. 2016b. The categorial status of the pronoun ku [in Korean]. Studies in Modern Grammar 87:45–58.

Kim, Young-Kook. 1997b. Agreement phrases in DP. University College London Working Papers in Linguistics 9:1–24.

Ko, Heejeong. 2007. Asymmetries in scrambling and cyclic linearization. Linguistic Inquiry 38:49–83.

Koak, Heeshin. 2008. A morpho-syntactic approach to pronominal binding. University of Pennsylvania Working Papers in Linguistics 14:227–240.

Koeneman, Olaf Nicolaas Cornelis Johannes. 2000. The ﬂexible nature of verb movement. Doctoral Dissertation, Utrecht University, Utrecht, the Netherlands.

Koizumi, Masatoshi. 1994. Secondary predicates. Journal of East Asian Linguistics 3:25–79.

Koulidobrova, Elena, and Diane Lillo-Martin. 2016. A ‘point’ of inquiry: The case of the (non-)pronominalIX in ASL. In The impact of pronominal form on interpretation, ed. Patrick Grosz and Pritty Patel-Grosz, 221–250. Berlin: Mouton de Gruyter.

Kuroda, Sige-Yuki. 1965. Generative grammatical studies in the Japanese language. Doc- toral Dissertation, MIT, Cambridge, Massachusettes.

Kweon, Soo-Ok. 2017. L1/l2 inﬂuence in the interpretation of pronouns. Foreign Languages Education 24:23–42.

Kwon, Nayoung, Maria Polinsky, Robert Kluender, and Miseon Lee. 2009. Anaphoric inventories and bound variable interpretation: Evidence from Korean. Poster presented at the 22nd Annual CUNY Conference on Human Sentence Processing, University of California, Davis.

LaCara, Nicholas. 2010. Verbal ellipsis in the nominal domain. Master’s Thesis, University of California, Santa Cruz.

Lai, Sandra. 1998. The grammar and acquisition of Secwepemctsín independent pronouns. Master’s Thesis, University of British Columbia, Vancouver, Canada.

Laka, Mugarza. 1996. A brief grammar of Euskara, the Basque language. Unpublished manuscript, University of the Basque Country, Vitoria-Gasteiz, Spain.

Lappin, Shalom. 1996. The interpretation of ellipsis. In The handbook of contemporary semantic, ed. Shalom Lappin, 145–175. Oxford, UK: Blackwell Publishing.

Lasnik, Howard. 1999. Minimalist analysis. Oxford, UK: Blackwell Publishing.

Lasnik, Howard. 2003. Minimalist investigations in linguistic theory. London and New York: Routledge.

Lasnik, Howard, and Tim Stowell. 1991. Weakest crossover. Linguistic Inquiry 22:687–720.

Lee, Chang-Wook. 2005a. Caki revisited. Study of English linguistics and English literature 9:199–215.

154 Lee, Eun-Ji. 2005b. The status of the null object in the VP ellipsis context. Studies in Generative Grammar 15:609–629.

Lee, Eunsuk. 2008. A single restriction on scrambling in Korean. Linguistics 16:73–90.

Lee, Ik-Sup, Sang-Uk Lee, and Wan Chae. 1997. Korean language [in Korean]. Seoul: Singu Mwunhwasa.

Lee, Kimoon. 1978. Korean personal pronouns [in Korean]. Kwanakemwunyenkwu 3:325– 338.

Lee, Wooseung. 2016. Argument ellipsis vs. V-stranding VP ellipsis in Korean: Evidence from disjunction. Language Research 33:1–20.

Lee, Wooseung, and Jihyun Kim. 2010. DP ellipsis as independent phenomena from pro in pro-drop languages. Korean Journal of Linguistics 35:1009–1029.

Li, Grace. 1998. Null object and VP ellipsis in Chinese. In Proceedings of the 9th North American Conference on Chinese Linguistics, ed. Lin Hua, 155–172. Los Angeles, Cal- ifornia: GSIL, University of Southern California.

Liejiong, Xu. 1986. Free empty category. Linguistic Inquiry 17:75–93.

Lim, Kihong. 1998. A split analysis of caki-binding in Korean. Doctoral Dissertation, University of Hawai’i, M¯anoa.

Madigan, Sean. 2015. Anaphora and binding. In The handbook of Korean linguistics, ed. Lucien Brown and Jaehoon Yeon, 137–156. Chichester, UK: John Wiley and Sons.

Marsden, Heather. 2012. Poverty of the stimulus and UG in Japanese L2 acquisition: A response to Sheen (2000). Durham Working Papers in Linguistics 8:67–80.

Massam, Diane. 1992. Null objects and non-thematic subjects. Journal of Linguistics 28:115–137.

Massam, Diane, and Yves Roberge. 1989. Recipe context null objects in English. Linguistic Inquiry 20:134–139.

May, Robert. 1977. The grammar of quantiﬁcation. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

May, Robert. 1985. Logical form: Its structure and derivation. Cambridge, Massachusettes: MIT Press.

McCloskey, James. 1991. Clause structure, ellipsis and proper government in Irish. Lingua 85:259–302.

McKillen, Alanah. 2016. On the interpretation of reﬂexive pronouns. Doctoral Dissertation, McGill University.

Merchant, Jason. 2001. The syntax of silence: Sluicing, islands, and the theory of ellipsis. Oxford, UK: Oxford University Press.

155 Merchant, Jason. 2013a. Diagnosing ellipsis. In Diagnosing syntax, ed. Lisa Lai-Shen Cheng and Norbert Corver, 485–501. Oxford, UK: Oxford University Press.

Merchant, Jason. 2013b. Yet another look at deep and surface anaphora. Unpublished manuscript, University of Chicago.

Miyamoto, Edson T. 2006. Processing alternative word orders in Japanese. In Hand- book of East Asian psycholinguistics, vol. 2: Japanese, ed. Mineharu Nakayama, Reiko Mazuka, and Yasuhiro Shirai, 257–263. Cambridge, UK: Cambridge University Press.

Montalbetti, Mario. 1984. After binding: On the interpretation of pronouns. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Moon, Gui-Sun. 2010. Null arguments redux. The Linguistic Association of Korea Journal 18:67–92.

Murphy, Gregory L. 1985. Psychological explanations of deep and surface anaphora. Journal of Pragmatics 9:785–813.

Neeleman, Ad, and Kriszta Szendrői. 2007. Radical pro drop and the morphology of pronouns. Linguistic Inquiry 38:671–714.

Nishigauchi, Taisuke. 1990. Quantiﬁcation in the theory of grammar. Dordrecht: Kluwer Academic Publishers.

Noguchi, Tohru. 1997. Two types of pronouns and variable binding. Language 770–797.

O’Grady, William. 1984. The syntax of Korean anaphora. Language Research 20:121–138.

Oh, Sun-Young. 2009. Invoking categories through co-present person reference: The case of Korean conversation. Journal of Pragmatics 42:1219–1242.

Ohtaki, Koichi. 2011. Argument ellipsis arising from non-fusional case morphology. In Online Proceedings of GLOW in Asia Workshop for Young Scholars, ed. Koichi Otaki, Hajime Takeyasu, and Shin-ichi Tanigawa, 247–261. Mie, Japan: Mie University.

Ohtaki, Koichi. 2014. Ellipsis of arguments: Its acquisition and theoretical implications. Doctoral Dissertation, University of Connecticut, Storrs.

Oku, Satoshi. 1998. A theory of selection and reconstruction in the minimalist perspective. Doctoral Dissertation, University of Connecticut, Storrs.

Oosterhof, Albert. 2008. The semantics of generics in Dutch and related languages. Ams- terdam: John Benjamins Publishing.

Otani, Kazuyo, and John Whitman. 1991. V-raising and VP-ellipsis. Linguistic Inquiry 345–358.

Pan, Haihua. 2002. Null object constructions, VP-ellipsis, and sentence interpretation. In Linguistics by heart: Webfest for horst-dieter gasde, berlin: Zas, ed. Daniel Hole, Paul Law, and Niina Zhang, 1–9. Berlin: ZAS.

Panagiotidis, Phoevos. 2002. Pronouns, clitics and empty nouns: ‘Pronominality’ and licensing in syntax. Amsterdam: John Benjamins Publishing.

156 Park, Myung-Kwan. 1994. A morpho-syntactic study of Korean verbal inﬂection. Doctoral Dissertation, University of Connecticut, Storrs.

Park, Myung-Kwan. 1997. The syntax of VP ellipsis in Korean. Language Research 33:629– 648.

Park, Myung Kwan. 2013. Debunking null objects in Korean: From pro to DP/NP ellipsis. Studies in Modern Grammar 72:41–66.

Park, Myung Kwan. 2015. Extraction out of overt anaphora: Korean kule(h). Linguistic Research 32:693–718.

Peirce, Jonathan W. 2007. Psychopy–psychophysics software in Python. Journal of Neu- roscience Methods 162:8–13.

Peirce, Jonathan W. 2009. Generating stimuli for neuroscience using PsychoPy. Frontiers in Neuroinformatics 2:10.

Pérez-Leroux, Ana T, and William R Glass. 1999. Null anaphora in Spanish second language acquisition: probabilistic versus generative approaches. Second Language Research 15:220–249.

Perlmutter, David. 1971. Deep and surface structure constraints in syntax. New York: Holt, Rinehart and Wilson.

Picallo, Carme. 1994. Catalan possessive pronouns: The Avoid Pronoun Principle revisited. Natural Language and Linguistic Theory 12:259–299.

Pollard, Carl, and Ivan Sag. 1994. Head-driven phrase structure grammar. Chicago: Uni- versity of Chicago Press.

Postal, Paul. 1969. On so-called pronouns in English. Monograph Series on Language and Linguistics 19:177–206.

Postal, Paul. 1972. Some further limitations of interpretive theories of anaphora. Linguistic Inquiry 3:349–371.

Reinhart, Tanya. 1983. Anaphora and semantic interpretation. Chicago: University of Chicago Press.

Reinhart, Tanya. 1986. Center and periphery in the grammar of anaphora. In Studies in the acquisition of anaphora, vol. 1 , ed. Babara Lust, 123–150. Dordrecht, the Netherlands: Reidel Publishing Company.

Rizzi, Luigi. 1982. Issues in Italian syntax. Berlin: Mouton de Gruyter.

Rizzi, Luigi. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry 17:501– 557.

Roberts, Ian, and Anders Holmberg. 2010. Introduction: Parameters in minimalist theory. In Parametric variation: Null subjects in minimalist theory, ed. Theresa Biberauer, Anders Holmberg, Ian Roberts, and Michelle Sheehan, 1–57. Cambridge, UK: Cam- bridge University Press.

157 Ross, John. 1967. Constraints on variables in syntax. Doctoral Dissertation, MIT, Cam- bridge, Massachusettes.

Saﬁr, Ken. 1999. Vehicle change and reconstruction in A-chains.¯ Linguistic Inquiry 30:587– 620.

Sag, Ivan A. 1976. Deletion and logical form. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Sag, Ivan A, and Jorge Hankamer. 1984. Toward a theory of anaphoric processing. Lin- guistics and Philosophy 7:325–345.

Saito, Mamoru. 1985. Some asymmetries in Japanese and their theoretical implications. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Saito, Mamoru. 2003. On the role of selection in the application of merge. In Proceedings of the 33rd Annual Meeting of the North East Linguistic Society, ed. Shigeto Kawa- hara, and Makoto Kadowaki, 323–345. Amherst, Massachusettes.: GLSA, University of Massachusettes, Amherst.

Saito, Mamoru. 2007. Notes on East Asian argument ellipsis. Language Research 43:203– 227.

Saito, Mamoru, and Duk-Ho An. 2010. A comparative syntax of ellipsis in Japanese and Korean. In Proceedings of the 6th Workshop on Altaic Formal Linguistics, 287–307. Cambridge, Massachusettes: Working Papers in Linguistics.

Sakamoto, Yuta. 2015. Disjunction as a new diagnostic for (argument) ellipsis. In Pro- ceedings of the 45th Annual Meeting of North East Linguistic Society, 15–28. Amherst, Massachusettes: GLSA, University of Massachusettes, Amherst.

Sakamoto, Yuta. 2016. Phases and argument ellipsis in Japanese. Journal of East Asian Linguistics 25:243–274.

Sato, Yosuke. 2014. Argument ellipsis in Colloquial Singapore English and the anti- agreement hypothesis. Journal of Linguistics 50:365–401.

Sato, Yosuke. 2016. Remarks on the parameters of argument ellipsis: A new perspective from Colloquial Singapore English. Syntax 19:392–411.

Sato, Yosuke, and Simin Karimi. 2016. Subject-object asymmetries in Persian argument ellipsis and the anti-agreement theory. Glossa: A Journal of General Linguistics 1:1–31.

Schøsler, Lene. 1994. Feature analysis of Danish pronominal paradigms with a view to a Danish application of the pronominal approach. International Journal of Lexicography 7:118–141.

Schuyler, Tamara. 2001. Wh-movement out of the site of VP ellipsis. Syntax and semantics at Santa Cruz 3:1–20.

Şener, Serkan, and Daiko Takahashi. 2010. Ellipsis of arguments in Japanese and Turkish. Nanzan Linguistics 6:79–99.

158 Shaﬁei, Nazila. 2015. Ellipsis in Persian complex predicates: VVPE or something else. In Proceedings of the 2015 Annual Conference of the Canadian Linguistics Association.

Shim, Bongsup. 1993. Comparative study of Ku, CELUI-CI, IL (From deixis to anaphora) [in Korean]. Studies in Generative Grammar 3:221–243.

Shim, Ji-Young, and Marcel Den Dikken. 2007. The tense of resultatives: The case of Korean. Unpublished manuscript, CUNY Graduate Center [presented at JK 18 and NELS 38].

Shinohara, Michie. 2006. On some diﬀerences between the major deletion phenomena and Japanese argument ellipsis. Unpublished manuscript, Nanzan University, Nagoya, Japan.

Simpson, Andrew, Arunima Choudhury, and Mythili Menon. 2013. Argument ellipsis and the licensing of covert nominals in Bangla, Hindi and Malayalam. Lingua 134:103–128.

Sobin, Nicholas. 2008. Do so and VP. Linguistic Inquiry 39:147–160.

Sohn, Ho-Min. 1994. Korean. London: Routledge.

Sohn, Ho-Min. 1999. The Korean Language. Cambridge, UK: Cambridge University Press.

Son, Minjeong. 2006. Directed motion and non-predicative pathP in Korean. In Tromsø Working Papers in Language and Linguistics 33:2, special issue on adpositions, ed. Pete Svenoniusr, 176–199. Tromsø, Norway: CASTL, University of Tromsø.

Song, Hee-Jeong. 2013. Second language acquisition of pronominal binding by learners of Korean and English. Doctoral Dissertation, University of Southampton, UK.

Speas, Margaret. 2004. Economy, agreement and the representation of null arguments. Unpublished manuscript, University of Massachusetts, Amherst.

Storoshenko, Dennis Ryan. 2008. A bound-variable account of the Korean anaphor caki. In Proceedings of the 26th West Coast Conference on Formal Linguistics, ed. Charles B. Chang and Hannah J. Haynie, 438–444. Somerville, Massachusettes: Cascadilla Pro- ceedings Project.

Stroik, Thomas. 2001. On the light verb hypothesis. Linguistic Inquiry 32:362–369.

Suh, Jinhee. 1990. Scope phenomena and aspects of Korean syntax. Doctoral Dissertation, University of Southern California.

Takahashi, Daiko. 1994. Sluicing in Japanese. Journal of East Asian Linguistics 3:265–300.

Takahashi, Daiko. 2007. Argument ellipsis from a crosslinguistic perspective: An interim report. Paper presented at GLOW in Asia 6.

Takahashi, Daiko. 2008. Quantiﬁcational null objects and argument ellipsis. Linguistic Inquiry 39:307–326.

Takahashi, Daiko. 2013. Argument ellipsis in Japanese and Malayalam. Nanzan Linguistics 9:173–192.

159 Takita, Kensuke. 2011. An argument for argument ellipsis from -sika NPIs. In Proceedings of the 39th Annual Meeting of North East Linguistic Society, ed. Suzi Lima, Kevin Mullin, and Brian Smith, 771–784. Amherst, Massachusettes: GLSA, University of Massachusettes, Amherst.

Tanaka, Hidekazu. 2008. Clausal complement ellipsis. Unpublished manuscript, University of York, UK.

Tancredi, Chris. 1992. Deletion, deaccenting, and presupposition. Doctoral Dissertation, MIT, Cambridge, Massachusettes.

Tanenhaus, Michael, and Greg Carlson. 1990. Comprehension of deep and surface verb phrase anaphors. Language and Cognitive Processes 5:257–280.

Taraldsen, Tarald. 1978. On the NIC, vacuous application and the that-trace ﬁlter. Unpub- lished manuscript, MIT, Cambridge, Massachusettes.

Tayyebi, Ghasem, Firooz Sadighi, and Morteza Yamini. 2011. The availability of universal grammar to second language learners: A case of overt pronoun constraint. In 2011 2nd International Conference on Humanities, Historical and Social Sciences, 221–225. Singapore: IACSIT Press.

Tomioka, Satoshi. 2003. The semantics of Japanese null pronouns and its cross-linguistic implications. In The interfaces: Deriving and interpreting omitted structures, ed. Kerstin Schwabe and Suzanne Winkler, 321–339. Amsterdam: John Benjamins Publishing.

Toosarvandani, Maziar. 2009. Ellipsis in Farsi complex predicates. Syntax 12:60–92.

Um, Hong-Joon. 2011. The nature of the null arguments in Korean. Studies in Modern Grammar 63:73–93.

Ward, Gregory, Richard Sproat, and Gail McKoon. 1991. A pragmatic analysis of so-called anaphoric islands. Language 67:439–472.

Wasow, Thomas. 1972. Anaphoric relations in English. Doctoral Dissertation, MIT, Cam- bridge, Massachusettes.

Wasow, Thomas. 1986. Reﬂections on anaphora. In Studies in the acquisition of anaphora: Volume I: Deﬁning the constraints, ed. Barbara Lust, 107–122. Dordrecht, the Nether- lands: Reidel Publishing Company.

Wei, Ting-Chi, and Yen-hui Audrey Li. 2016. How to do so in Mandarin Chinese. Journal of East Asian Linguistics 25:183–212.

White, Lydia. 2003. Second language acquisition and Universal Grammar. Cambridge, UK: Cambridge University Press.

Wilder, Chris. 1997. Some properties of ellipsis in coordination. In Studies on universal grammar and typological variation, ed. Artemis Alexiadou and Tracy Hall, 59–107. Amsterdam: John Benjamins Publishing.

Williams, Edwin S. 1977. On “Deep and Surface Anaphora”. Linguistic Inquiry 8:692–696.

160 Wiltschko, Martina. 1998. On the syntax and semantics of (relative) pronouns and determiners. The Journal of Comparative Germanic Linguistics 2:143–181.

Yashima, Jun. 2015. On the apparent unbindability of overt third-person pronouns in Japanese. Natural Language and Linguistic Theory 33:1421–1438.

Yoon, James. 1985. On the treatment of empty categories in topic prominent languages. Unpublished manuscript, University of Illinois, Urbana-Champaign.

Yu, Gyeong-Min. 2005. A diachronic study of the Korean demonstrative jeo [in Korean]. Korean Semantics 16:179–204.

161 Appendix A

Test sentences from Experiment 1A

A.1 Condition 1: Quantiﬁcational-Overt

(1) Hanswu, Cinswu, Minswu-ka nongkwucang-eyse nongkwu-lul ha-ko Hanswu, Cinswu, Minswu-nom basketball.court-at basketball-acc do-prog iss-ess-ta. Hanswu-ka caki-ka nongkwu-lul cal ha-n-tako past-decl Hanswu-nom self-nom basketball-acc well do-pres-comp malha-yess-ta. Cinswu-to caki-ka nongkwu-lul cal ha-n-tako say-past-decl Cinswu-also self-nom basketball-acc well do-pres-comp malha-yess-ta. Minswu-to caki-ka nongkwu-lul cal ha-n-tako say-past-decl Minswu-also self-nom basketball-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu said that he plays basketball well. Cinswu also said that he plays basketball well. Minswu also said that he plays basketball well.’ Motwu-ka nongkwucang-eyse ku-ka nongkwu-lul cal everyone-nom basketball.court-at he-nom basketball-acc well ha-n-tako malha-yess-ta. do-pres-comp say-past-decl ‘Everyone said at a basketball court that he plays basketball well.’

162 cengmal coaha-n-tako malha-yess-ta. Sengswu-to caki-ka really like-pres-comp say-past-decl Sengswu-also self-nom maikhulcayksun-ul cengmal coaha-n-tako malha-yess-ta. Michael.Jackson-acc really like-pres-comp say-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a music café. Wonmwu said that he really likes Michael Jackson. Kiswu also said that he really likes Michael Jackson. Sengswu also said that he really likes Michael Jackson.’ Motwu-ka mywucikkkaphey-eyse ku-ka maikhulcayksun-ul everyone-nom music.café-at he-nom Michael.Jackson-acc cengmal coaha-n-tako malha-yess-ta. really like-pres-comp say-past-decl ‘Everyone said at a music café that he really likes Michael Jackson.’

(3) Hwanswu, Thayswu, Twusik-ika nolaypang-eyse nol-ko iss-ess-ta. Hwanswu, Thayswu, Twusik-nom singing.room-at hang.out-prog past-decl Hwanswu-ka caki-ka pillicin-ul cal pwulu-n-tako malha-yess-ta. Hwanswu-nom self-nom Billie.Jean-acc well sing-pres-comp say-past-decl Thayswu-to caki-ka pillicin-ul cal pwulu-n-tako malha-yess-ta. Thayswu-also self-nom Billie.Jean-acc well sing-pres-comp say-past-decl Twusik-ito caki-ka pillicin-ul cal pwulu-n-tako malha-yess-ta. Twusik-also self-nom Billie.Jean-acc well sing-pres-comp say-past-decl ‘Hwanswu, Thayswu, and Twusik were hanging out at a singing room. Hwanswu said that he sings Billie Jean well. Thayswu also said that he sings Billie Jean well. Twusik also said that he sings Billie Jean well.’ Motwu-ka nolaypang-eyse ku-ka pillicin-ul cal pwulu-n-tako everyone-nom singing.room-at he-nom Billie.Jean-acc well sing-pres-comp malha-yess-ta. say-past-decl ‘Everyone said at a singing room that he sings Billie Jean well.’

(4) Cinkwu, Minkwu, Sinswu-ka tosekwuan-eyse sihemkongpwu-lul ha-ko Cinkwu, Minkwu, Sinswu-nom library-in exam.studying-acc do-prog iss-ess-ta. Cinkwu-ka caki-ka swuhak-ul cal ha-n-tako past-decl Cinkwu-nom self-nom math-acc well do-pres-comp malha-yess-ta. Minkwu-to caki-ka swuhak-ul cal ha-n-tako say-past-decl Minkwu-also self-nom math-acc well do-pres-comp malha-yess-ta. Sinswu-to caki-ka swuhak-ul cal ha-n-tako say-past-decl Sinswu-also self-nom math-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Cinkwu, Minkwu, and Sinswu were studying for the exam in a library. Cinkwu said that he is good at math. Minkwu also said that he is good at math. Sinswu also said that he is good at math.’

163 Motwu-ka tosekwuan-eyse ku-ka swuhak-ul cal ha-n-tako everyone-nom library-in he-nom math-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Everyone said in a library that he is good at math.’

A.2 Condition 2: Quantiﬁcational-Null

(1) Hanswu, Cinswu, Minswu-ka nongkwucang-eyse nongkwu-lul ha-ko Hanswu, Cinswu, Minswu-nom basketball.court-at basketball-acc do-prog iss-ess-ta. Hanswu-ka caki-ka nongkwu-lul cal ha-n-tako past-decl Hanswu-nom self-nom basketball-acc well do-pres-comp malha-yess-ta. Cinswu-to caki-ka nongkwu-lul cal ha-n-tako say-past-decl Cinswu-also self-nom basketball-acc well do-pres-comp malha-yess-ta. Minswu-to caki-ka nongkwu-lul cal ha-n-tako say-past-decl Minswu-also self-nom basketball-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu said that he plays basketball well. Cinswu also said that he plays basketball well. Minswu also said that he plays basketball well.’ Motwu-ka nongkwucang-eyse pro nongkwu-lul cal everyone-nom basketball.court-at basketball-acc well ha-n-tako malha-yess-ta. do-pres-comp say-past-decl ‘Everyone said at a basketball court that he plays basketball well.’

(2) Wonmwu, Kiswu, Sengswu-ka mywucikkkaphey-eyse tayhwua-lul Wonmwu, Kiswu, Sengswu-nom music.café-at conversation-acc nanwu-ko iss-ess-ta. Wonmwu-ka caki-ka maikhulcayksun-ul cengmal share-prog past-decl Wonmwu-nom self-nom Michael.Jackson-acc really coaha-n-tako malha-yess-ta. Kiswu-to caki-ka maikhulcayksun-ul like-pres-comp say-past-decl Kiswu-also self-nom Michael.Jackson-acc cengmal coaha-n-tako malha-yess-ta. Sengswu-to caki-ka really like-pres-comp say-past-decl Sengswu-also self-nom maikhulcayksun-ul cengmal coaha-n-tako malha-yess-ta. Michael.Jackson-acc really like-pres-comp say-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a music café. Wonmwu said that he really likes Michael Jackson. Kiswu also said that he really likes Michael Jackson. Sengswu also said that he really likes Michael Jackson.’ Motwu-ka mywucikkkaphey-eyse pro maikhulcayksun-ul everyone-nom music.café-at Michael.Jackson-acc cengmal coaha-n-tako malha-yess-ta. really like-pres-comp say-past-decl

164 ‘Everyone said at a music café that he really likes Michael Jackson.’

(3) Hwanswu, Thayswu, Twusik-ika nolaypang-eyse nol-ko iss-ess-ta. Hwanswu, Thayswu, Twusik-nom singing.room-at hang.out-prog past-decl Hwanswu-ka caki-ka pillicin-ul cal pwulu-n-tako malha-yess-ta. Hwanswu-nom self-nom Billie.Jean-acc well sing-pres-comp say-past-decl Thayswu-to caki-ka pillicin-ul cal pwulu-n-tako malha-yess-ta. Thayswu-also self-nom Billie.Jean-acc well sing-pres-comp say-past-decl Twusik-ito caki-ka pillicin-ul cal pwulu-n-tako malha-yess-ta. Twusik-also self-nom Billie.Jean-acc well sing-pres-comp say-past-decl ‘Hwanswu, Thayswu, and Twusik were hanging out at a singing room. Hwanswu said that he sings Billie Jean well. Thayswu also said that he sings Billie Jean well. Twusik also said that he sings Billie Jean well.’ Motwu-ka nolaypang-eyse pro pillicin-ul cal pwulu-n-tako everyone-nom singing.room-at Billie.Jean-acc well sing-pres-comp malha-yess-ta. say-past-decl ‘Everyone said at a singing room that he sings Billie Jean well.’

165 A.3 Condition 3: Referential-Overt

166 (4) Cinkwu, Minkwu, Sinswu-ka tosekwuan-eyse sihemkongpwu-lul ha-ko Cinkwu, Minkwu, Sinswu-nom library-in exam.studying-acc do-prog iss-ess-ta. Cinkwu-ka caki-ka swuhak-ul cal ha-n-tako past-decl Cinkwu-nom self-nom math-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Cinkwu, Minkwu, and Sinswu were studying for the exam in a library. Cinkwu said that he is good at math.’ Cinkwu-ka tosekwuan-eyse ku-ka swuhak-ul cal ha-n-tako Cinkwu-nom library-in he-nom math-acc well do-pres-comp malha-yess-ta. say-past-decl ‘Cinkwu said in a library that he is good at math.’

A.4 Condition 4: Referential-Null

167 ‘Wonmwu said at a music café that he really likes Michael Jackson.’

168 Appendix B

Test sentences from Experiment 1B

B.1 Condition 1: Simple

(1) Hanswu, Cinswu, Minswu-ka nongkwucang-eyse nongkwu-lul ha-ko Hanswu, Cinswu, Minswu-nom basketball.court-at basketball-acc do-prog iss-ess-ta. Hanswu-ka caki-uy umlyoswu-lul masi-ess-ta. Cinswu-to -past-decl Hanswu-nom self-gen beverage-acc drink-past-decl Cinswu-also caki-uy umlyoswu-lul masi-ess-ta. Minswu-to caki-uy umlyoswu-lul self-gen beverage-acc drink-past-decl Minswu-also self-gen beverage-acc masi-ess-ta. drink-past-decl ‘Hanswu, Cinswu, and Minswu were playing basketball at a basketball court. Hanswu drank his own beverage. Cinswu also drank his own beverage. Minswu also drank his own beverage.’ Motwu-ka nongkwucang-eyse ku-uy umlyoswu-lul masi-ess-ta. everyone-nom basketball.court-at he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage at a basketball court.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse yakwu-lul ha-ko Yonghi Kenhi Myengho-nom play.ground-in baseball-acc do-prog iss-ess-ta. Yonghi-ka caki-uy kong-ul tenci-ess-ta. Kenhi-to past-decl Yonghi-nom self-gen ball-acc throw-past-decl Kenhi-also caki-uy kong-ul tenci-ess-ta. Myengho-to caki-uy kong-ul self-gen ball-acc throw-past-decl Myengho-also self-gen ball-acc tenci-ess-ta. throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground. Yonghi threw his own ball. Kenhi also threw his own ball. Myengho also threw his own ball.’ Motwu-ka wuntongcang-eyse ku-uy kong-ul tenci-ess-ta. everyone-nom play.ground-in he-gen ball-acc throw-past-decl

169 ‘Everyone throw his ball in a playground.’

(3) Kwangho, Changho, Yengho-ka siktang-eyse cenyek-ul mek-ko Kwangho Changho Yengho-nom restaurant-at dinner-acc eat-prog iss-ess-ta. Kwangho-ka caki-uy suphakeythi-lul mek-ess-ta. past-decl Kwangho-nom self-gen spaghetti-acc eat-past-decl Changho-to caki-uy suphakeythi-lul mek-ess-ta. Yengho-to caki-uy Changho-also self-gen spaghetti-acc eat-past-decl Yengho-also self-gen suphakeythi-lul mek-ess-ta. spaghetti-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having dinner at a restaurant. Kwangho ate his own spaghetti. Changho also ate his own spaghetti. Yengho also ate his own spaghetti.’ Motwu-ka siktang-eyse ku-uy suphakeythi-lul mek-ess-ta. everyone-nom restaurant-at he-gen spaghetti-acc eat-past-decl ‘Everyone ate his spaghetti at a restaurant.’

(4) Chiswu, Pyengswu, Hanswu-ka chatcip-eyse tayhwua-lul nanwu-ko Chiswu Pyengswu Hanswu-nom tea.house-at conversation-acc share-prog iss-ess-ta. Chiswu-ka caki-uy hyeng-ul helttut-ess-ta. past-decl Chiswu-nom self-gen older.brother-acc speak.ill.of-past-decl Pyengswu-to caki-uy hyeng-ul helttut-ess-ta. Hanswu-to Pyengswu-also self-gen older.brother-acc speak.ill.of-past-decl Hanswu-also caki-uy hyeng-ul helttut-ess-ta. self-gen older.brother-acc speak.ill.of-past-decl ‘Chiswu, Pyengswu, and Hanswu were having a conversation at a tea house. Chiswu spoke ill of his own older brother. Pyengswu also spoke ill of his own older brother. Hanswu also spoke ill of his own older brother.’ Motwu-ka chatcip-eyse ku-uy hyeng-ul helttut-ess-ta. everyone-nom tea.house-at he-gen older.brother-acc speak.ill.of-past-decl ‘Everyone spoke ill of his own older brother at a tea house.’

B.2 Condition 2: Complex

The test items for the Complex condition are the same as those for the Quantificational- Overt condition in Experiment 1A.

170 Appendix C

Test sentences from Experiment 2

C.1 Condition 1: Bound-August

The test items for the Bound-August condition are the same as those for the Simple condition in Experiment 1B.

C.2 Condition 2: Bound-September

(1) Chiswu, Pyengswu, Hanswu-ka kkaphey-eyse tayhwua-lul nanwu-ko Chiswu Pyengswu Hanswu-nom café-at conversation-acc share-prog iss-ess-ta. Chiswu-ka Chiswu-uy cen yecachinkwu-lul past-decl Chiswu-nom Chiswu-gen previous girlfriend-acc helttut-ess-ta. Pyengswu-to Pyengswu-uy cen yecachinkwu-lul speak.ill.of-past-decl Pyengswu-also Pyengswu-gen previous girlfriend-acc helttut-ess-ta. Hanswu-to Hanswu-uy cen yecachinkwu-lul speak.ill.of-past-decl Hanswu-also Hanswu-gen previous girlfriend-acc helttut-ess-ta. speak.ill.of-past-decl ‘Chiswu, Pyengswu, and Hanswu were having a conversation at a café. Chiswu spoke ill of Chiswu’s ex-girlfriend. Pyengswu also spoke ill of Pyengswu’s ex- girlfriend. Hanswu also spoke ill of Hanswu’s ex-girlfriend.’ Motwu-ka ku-uy cen yecachinkwu-lul helttut-ess-ta. everyone-nom he-gen previous girlfriend-acc speak.ill.of-past-decl ‘Everyone spoke ill of his ex-girlfriend.’

(2) Congswu, Hwuanswu, Tongswu-ka kongwuen-eyse nol-ko Congswu Hwuanswu Tongswu-nom park-at hang.out-prog iss-ess-ta. Congswu-ka Congswu-uy kay-lul kancilephi-ess-ta. past-decl Congswu-nom Congswu-gen dog-acc tickle-past-decl

171 Hwuanswu-to Hwuanswu-uy kay-lul kancilephi-ess-ta. Tongswu-to Hwuanswu-also Hwuanswu-gen dog-acc tickle-past-decl Tongswu-also Tongswu-uy kay-lul kancilephi-ess-ta. Tongswu-gen dog-acc tickle-past-decl ‘Congswu, Hwuanswu, and Tongswu were hanging out at a park. Congswu tickled Congswu’s dog. Hwuanswu also tickled Hwuanswu’s dog. Tongswu also tickled Tongswu’s dog.’ Motwu-ka ku-uy kay-lul kancilephi-ess-ta. everyone-nom he-gen dog-acc tickle-past-decl ‘Everyone tickled his dog.’

(3) Tongwu, Cinwu, Thaywu-ka Sinswu-uy cip-eyse nol-ko Tongwu Cinwu Thaywu-nom Sinswu-gen house-at hang.out-prog iss-ess-ta. Tongwu-ka Tongwu-uy yetongsayng-ul ttayli-ess-ta. past-decl Tongwu-nom Tongwu-gen younger.sister-acc hit-past-decl Cinwu-to Cinwu-uy yetongsayng-ul ttayli-ess-ta. Thaywu-to Cinwu-also Cinwu-gen younger.sister-acc hit-past-decl Thaywu-also Thaywu-uy yetongsayng-ul ttayli-ess-ta. Thaywu-gen younger.sister-acc hit-past-decl ‘Tongwu, Cinwu, and Thaywu were hanging out at Sinswu’s house. Tongwu hit Tongwu’s younger sister. Cinwu also hit Cinwu’s younger sister. Thaywu also hit Thaywu’s younger sister.’ Motwu-ka ku-uy yetongsayng-ul ttayli-ess-ta. everyone-nom he-gen younger.sister-acc hit-past-decl ‘Everyone hit his younger sister.’

(4) Wonmwu, Kiswu, Sengswu-ka chatcip-eyse tayhwua-lul nanwu-ko Wonmwu Kiswu Sengswu-nom tea.house-at conversation-acc share-prog iss-ess-ta. Wonmwu-ka Wonmwu-uy nwuna-lul past-decl Wonmwu-nom Wonmwu-gen older.sister-acc chingchanha-yss-ta. Kiswu-to Kiswu-uy nwuna-lul compliment-past-decl Kiswu-also Kiswu-gen older.sister-acc chingchanha-yss-ta. Sengswu-to Sengswu-uy nwuna-lul compliment-past-decl Sengswu-also Sengswu-gen older.sister-acc chingchanha-yss-ta. compliment-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a tea house. Won- mwu complimented Wonmwu’s older sister. Kiswu also complimented Kiswu’s older sister. Sengswu also complimented Sengswu’s older sister.’ Motwu-ka ku-uy nwuna-lul chingchanha-yss-ta. everyone-nom he-gen older.sister-acc compliment-past-decl ‘Everyone complimented his older sister.’

172 C.3 Condition 3: Free-August

(1) Pyengmin, Cengmin, Tongmin-ika wuntongcang-eyse, Cengki-ka Pyengmin Cengmin Tongmin-nom play.ground-in Cengki-nom oki-lul kitalimye, chwukkwu-lul ha-ko iss-ess-ta. Pyengmin-ika coming-acc waiting soccer-acc do-prog past-decl Pyengmin-nom Cengki-uy kong-ul cha-ss-ta. Cengmin-ito Cengki-uy kong-ul Cengki-gen ball-acc kick-past-decl Cengmin-also Cengki-gen ball-acc cha-ss-ta. Tongmin-ito Cengki-uy kong-ul cha-ss-ta. kick-past-decl Tongmin-also Cengki-gen ball-acc kick-past-decl ‘Pyengmin, Cengmin, and Tongmin were playing soccer in a playground, waiting for Cengki to come. Pyengmin kicked Cengki’s ball. Cengmin also kicked Cengki’s ball. Tongmin also kicked Cengki’s ball.’ Motwu-ka ku-uy kong-ul cha-ss-ta. everyone-nom he-gen ball-acc kick-past-decl ‘Everyone kicked his ball.’

(2) Cinswu, Sungswu, Pyengswu-ka seythaksil-eyse, Cinho-ka oki-lul Cinswu Sungswu Pyengswu-nom laundry.room-at Cinho-nom coming-acc kitalimye, seythak-ul ha-ko iss-ess-ta. Cinswu-ka Cinho-uy waiting laundry-acc do-prog past-decl Cinswu-nom Cinho-gen seycey-lul sayongha-yss-ta. Sungswu-to Cinho-uy seycey-lul detergent-acc use-past-decl Sungswu-also Cinho-gen detergent-acc sayongha-yss-ta. Pyengswu-to Cinho-uy seycey-lul sayongha-yss-ta. use-past-decl Pyengswu-also Cinho-gen detergent-acc use-past-decl ‘Cinswu, Sungswu, and Pyengswu were doing laundry at a laundry room, waiting for Cinho to come. Cinswu used Cinho’s detergent. Sungswu also used Cinho’s detergent. Pyengswu also used Cinho’s detergent.’ Motwu-ka ku-uy seycey-lul sayongha-yss-ta. everyone-nom he-gen detergent-acc use-past-decl ‘Everyone used his detergent.’

(3) Kimkenmo, Kimcanghwun, Sinsunghwun-i nolaypang-eyse, Imwunsey-ka Kimkenmo Kimcanghwun Sinsunghwun-nom singing.room-at Imwunsey-nom oki-lul kitalimye, nol-ko iss-ess-ta. Kimkenmo-ka coming-acc waiting hang.out-prog past-decl Kimkenmo-nom Imwunsey-uy hithukok-ul pwulu-ess-ta. Kimcanghwun-to Imwunsey-gen hit.song-acc sing-past-decl Kimcanghwun-also Imwunsey-uy hithukok-ul pwulu-ess-ta. Sinsunghwun-to Imwunsey-uy Imwunsey-gen hit.song-acc sing-past-decl Sinsunghwun-also Imwunsey-gen hithukok-ul pwulu-ess-ta. hit.song-acc sing-past-decl ‘Kimkenmo, Kimcanghwun, and Sinsunghwun were hanging out in a singing room, waiting for Imwunsey to come. Kimkenmo sang Imwunsey’s hit song. Kim-

173 canghwun also sang Imwunsey’s hit song. Sinsunghwun also sang Imwunsey’s hit song.’1 Motwu-ka ku-uy hithukok-ul pwulu-ess-ta. everyone-nom he-gen hit.song-acc sing-past-decl ‘Everyone sang his hit song.’

(4) Pongswu, Thaykang, Thayho-ka pwuek-eyse, Sungki-ka oki-lul Pongswu Thaykang Thayho-nom kitchen-at Sungki-nom coming-acc kitalimye, selkeci-lul ha-ko iss-ess-ta. Pongswu-ka Sungki-uy waiting dish.washing-acc do-prog past-decl Pongswu-nom Sungki-gen khep-ul sis-ess-ta. Thaykang-ito Sungki-uy khep-ul cup-acc wash-past-decl Thaykang-also Sungki-gen cup-acc sis-ess-ta. Thayho-to Sungki-uy khep-ul sis-ess-ta. wash-past-decl Thayho-also Sungki-gen cup-acc wash-past-decl ‘Pongswu, Thaykang, and Thayho were doing the dishes at a kitchen, waiting for Sungki to come. Pongswu washed Sungki’s cup. Thaykang also washed Sungki’s cup. Thayho also washed Sungki’s cup.’ Motwu-ka ku-uy khep-ul sis-ess-ta. everyone-nom he-gen cup-acc wash-past-decl ‘Everyone washed his cup.’

C.4 Condition 4: Free-September

(1) Chiswu, Pyengswu, Hanswu-ka kkaphey-eyse, Cengki-ka oki-lul Chiswu Pyengswu Hanswu-nom café-at Cengki-nom coming-acc kitalimye, tayhwua-lul nanwu-ko iss-ess-ta. Chiswu-ka Cengki-uy waiting conversation-acc share-prog past-decl Chiswu-nom Cengki-gen cen yecachinkwu-lul helttut-ess-ta. Pyengswu-to Cengki-uy previous girlfriend-acc speak.ill.of-past-decl Pyengswu-also Cengki-gen cen yecachinkwu-lul helttut-ess-ta. Hanswu-to Cengki-uy previous girlfriend-acc speak.ill.of-past-decl Hanswu-also Cengki-gen cen yecachinkwu-lul helttut-ess-ta. previous girlfriend-acc speak.ill.of-past-decl ‘Chiswu, Pyengswu, and Hanswu were having a conversation at a café, waiting for Cengki to come. Chiswu spoke ill of Cengki’s ex-girlfriend. Pyengswu also spoke ill of Cengki’s ex-girlfriend. Hanswu also spoke ill of Cengki’s ex-girlfriend.’ Motwu-ka ku-uy cen yecachinkwu-lul helttut-ess-ta. everyone-nom he-gen previous girlfriend-acc speak.ill.of-past-decl ‘Everyone spoke ill of his ex-girlfriend.’ 1Kimkenmo, Kimcanghwun, Sinsunghwun, and Imwunsey are famous Korean male singers.

174 (2) Congswu, Hwuanswu, Tongswu-ka kongwuen-eyse, Cinkwu-ka oki-lul Congswu Hwuanswu Tongswu-nom park-at Cinkwu-nom coming-acc kitalimye, nol-ko iss-ess-ta. Congswu-ka Cinkwu-uy kay-lul waiting hang.out-prog past-decl Congswu-nom Cinkwu-gen dog-acc kancilephi-ess-ta. Hwuanswu-to Cinkwu-uy kay-lul kancilephi-ess-ta. tickle-past-decl Hwuanswu-also Cinkwu-gen dog-acc tickle-past-decl Tongswu-to Cinkwu-uy kay-lul kancilephi-ess-ta. Tongswu-also Cinkwu-gen dog-acc tickle-past-decl ‘Congswu, Hwuanswu, and Tongswu were hanging out at a park, waiting for Cinkwu to come. Congswu tickled Cinkwu’s dog. Hwuanswu also tickled Cinkwu’s dog. Tongswu also tickled Cinkwu’s dog.’ Motwu-ka ku-uy kay-lul kancilephi-ess-ta. everyone-nom he-gen dog-acc tickle-past-decl ‘Everyone tickled his dog.’

(3) Tongwu, Cinwu, Thaywu-ka Sinswu-uy cip-eyse, Sinswu-ka oki-lul Tongwu Cinwu Thaywu-nom Sinswu-gen house-at Sinswu-nom coming-acc kitalimye, nol-ko iss-ess-ta. Tongwu-ka Sinswu-uy waiting hang.out-prog past-decl Tongwu-nom Sinswu-gen yetongsayng-ul ttayli-ess-ta. Cinwu-to Sinswu-uy yetongsayng-ul younger.sister-acc hit-past-decl Cinwu-also Sinswu-gen younger.sister-acc ttayli-ess-ta. Thaywu-to Sinswu-uy yetongsayng-ul ttayli-ess-ta. hit-past-decl Thaywu-also Sinswu-gen younger.sister-acc hit-past-decl ‘Tongwu, Cinwu, and Thaywu were hanging out at Sinswu’s house, waiting for Sinswu to come. Tongwu hit Sinswu’s younger sister. Cinwu also hit Sinswu’s younger sister. Thaywu also hit Sinswu’s younger sister.’ Motwu-ka ku-uy yetongsayng-ul ttayli-ess-ta. everyone-nom he-gen younger.sister-acc hit-past-decl ‘Everyone hit his younger sister.’

(4) Wonmwu, Kiswu, Sengswu-ka chatcip-eyse, Tongwu-ka oki-lul Wonmwu Kiswu Sengswu-nom tea.house-at Tongwu-nom coming-acc kitalimye, tayhwua-lul nanwu-ko iss-ess-ta. Wonmwu-ka waiting conversation-acc share-prog past-decl Wonmwu-nom Tongwu-uy nwuna-lul chingchanha-yss-ta. Kiswu-to Tongwu-uy Tongwu-gen older.sister-acc compliment-past-decl Kiswu-also Tongwu-gen nwuna-lul chingchanha-yss-ta. Sengswu-to Tongwu-uy older.sister-acc compliment-past-decl Sengswu-also Tongwu-gen nwuna-lul chingchanha-yss-ta. older.sister-acc compliment-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a tea house, waiting for Tongwu to come. Wonmwu complimented Tongwu’s older sister. Kiswu also complimented Tongwu’s older sister. Sengswu also complimented Tongwu’s older sister.’

175 Motwu-ka ku-uy nwuna-lul chingchanha-yss-ta. everyone-nom he-gen older.sister-acc compliment-past-decl ‘Everyone complimented his older sister.’

176 Appendix D

Test sentences from Experiment 3

D.1 Condition 1: VPA-Bound (sloppy identity reading)

(1) Minswu, Kiswu, Cinswu-ka say kiswuksa-lo isaha-ko iss-ess-ta. Minswu Kiswu Cinswu-nom new dormitory-to move-prog past-decl Minswu-ka Minswu-uy cim-ul nalu-ess-ta. Kiswu-to Kiswu-uy Minswu-nom Minswu-gen stuff-acc move-past-decl Kiswu-also Kiswu-gen cim-ul nalu-ess-ta. stuff-acc move-past-decl ‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuff. Kiswu also moved Kiswu’s stuff.’ Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to kuleha-yess-ta. Minswu-nom he-gen stuff-acc move-past-conj Kiswu-also so.do-past-decl ‘Minswu moved his stuff, and Kiswu did so, too.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse yakwu-lul ha-ko Yonghi Kenhi Myengho-nom play.ground-in baseball-acc do-prog iss-ess-ta. Yonghi-ka Yonghi-uy kong-ul tenci-ess-ta. Kenhi-to past-decl Yonghi-nom Yonghi-gen ball-acc throw-past-decl Kenhi-also Kenhi-uy kong-ul tenci-ess-ta. Kenhi-gen ball-acc throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground. Yonghi threw Yonghi’s ball. Kenhi also threw Kenhi’s ball.’ Yonghi-ka ku-uy kong-ul tenci-ess-ko, Kenhi-to kuleha-yess-ta. Yonghi-nom he-gen ball-acc throw-past-conj Kenhi-also so.do-past-decl ‘Yonghi threw his ball, and Kenhi did so, too.’

177 (3) Thayswu, Kiswu, Cengswu-ka pa-eyse wuain-ul masi-ko iss-ess-ta. Thayswu Kiswu Cengswu-nom bar-at wuain-acc drink-prog past-decl Thayswu-ka Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-to Thayswu-nom Thayswu-gen wine.glass-acc break-past-decl Kiswu-also Kiswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-gen wine.glass-acc break-past-decl ‘Thayswu, Kiswu, and Cengswu were drinking wine at a bar. Thayswu broke Thayswu’s wine glass. Kiswu also broke Kiswu’s wine glass.’ Thayswu-ka ku-uy wuaincan-ul kkaythuli-ess-ko, Kiswu-to Thayswu-nom he-gen wine.glass-acc break-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl ‘Thayswu broke his wine glass, and Kiswu did so, too.’

(4) Myengswu, Hyenswu, Cwunswu-ka wuntong hwu swui-ko iss-ess-ta. Myengswu Hyenswu Cwunswu-nom exercise after rest-prog past-decl Myengswu-ka Myengswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-to Myengswu-nom Myengswu-gen beverage-acc drink-past-decl Hyenswu-also Hyenswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-gen beverage-acc drink-past-decl ‘Myengswu, Hyenswu, and Cwunswu were taking a rest after exercise. Myengswu drank Myengswu’s beverage. Hyenswu also drank Hyenswu’s beverage.’ Myengswu-ka ku-uy umlyoswu-lul masi-ess-ko, Hyenswu-to Myengswu-nom he-gen beverage-acc drink-past-conj Hyenswu-also kuleha-yess-ta. so.do-past-decl ‘Myengswu drank his beverage, and Hyenswu did so, too.’

(5) Pyengswu, Mwunswu, Cwunswu-ka cip-ul chengsoha-ko iss-ess-ta. Pyengswu Mwunswu Cwunswu-nom house-acc clean-prog past-decl Pyengswu-ka Pyengswu-uy pang-ul ssul-ess-ta. Mwunswu-to Pyengswu-nom Pyengswu-gen room-acc sweep-past-decl Mwunswu-also Mwunswu-uy pang-ul ssul-ess-ta. Mwunswu-gen room-acc sweep-past-decl ‘Pyengswu, Mwunswu, and Cwunswu were cleaning their house. Pyengswu swept Pyengswu’s room. Mwunswu also swept Mwunswu’s room.’ Pyengswu-ka ku-uy pang-ul ssul-ess-ko, Mwunswu-to Pyengswu-nom he-gen room-acc sweep-past-conj Mwunswu-also kuleha-yess-ta. so.do-past-decl ‘Pyengswu swept his room, and Mwunswu did so, too.’

178 (6) Chelswu, Yengswu, Cengswu-ka tosekwuan-eyse kongpwuha-ko iss-ess-ta. Chelswu Yengswu Cengswu-nom library-in study-prog past-decl Chelswu-ka Chelswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-to Chelswu-nom Chelswu-gen pen-acc drop-past-decl Yengswu-also Yengswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-gen pen-acc drop-past-decl ‘Chelswu, Yengswu, and Cengswu were studying in a library. Chelswu dropped Chelswu’s pen. Yengswu also dropped Yengswu’s pen.’ Chelswu-ka ku-uy pheyn-ul ttelethuli-ess-ko, Yengswu-to Chelswu-nom he-gen pen-acc drop-past-conj Chelswu-also kuleha-yess-ta. so.do-past-decl ‘Chelswu dropped his pen, and Yengswu did so, too.’

(7) Chelho, Phengho, Swuho-ka kakca phyenci-lul ssu-ko iss-ess-ta. Chelho Phengho Swuho-nom each letter-acc write-prog past-decl Chelho-ka Chelho-uy phyenci-lul ilk-ess-ta. Phengho-to Chelho-nom Chelho-gen letter-acc read-past-decl Phengho-also Phengho-uy phyenci-lul ilk-ess-ta. Phengho-gen letter-acc read-past-decl ‘Chelho, Phengho, and Swuho were each writing a letter. Chelho read Chelho’s letter. Phengho also read Phengho’s letter.’ Chelho-ka ku-uy phyenci-lul ilk-ess-ko, Phengho-to Chelho-nom he-gen letter-acc read-past-conj Phengho-also kuleha-yess-ta. so.do-past-decl ‘Chelho read his letter, and Phengho did so, too.’

(8) Kwangho, Changho, Yengho-ka ilsiktang-eyse cemsim-ul mek-ko Kwangho Changho Yengho-nom Japanese.restaurant-at lunch-acc eat-prog iss-ess-ta. Kwangho-ka Kwangho-uy utong-ul mek-ess-ta. past-decl Kwangho-nom Kwangho-gen Udon-acc eat-past-decl Changho-to Changho-uy utong-ul mek-ess-ta. Changho-also Changho-gen Udon-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having lunch at a Japanese restaurant. Kwangho ate Kwangho’s Udon. Changho also ate Changho’s Udon.’ Kwangho-ka ku-uy utong-ul mek-ess-ko, Changho-to Kwangho-nom he-gen Udon-acc eat-past-conj Changho-also kuleha-yess-ta. so.do-past-decl ‘Kwangho ate his Udon, and Changho did so, too.’

179 (9) Pyengmin, Cengmin, Tongmin-ika wuntongcang-eyse chwukkwu-lul ha-ko Pyengmin Cengmin Tongmin-nom play.ground-in soccer-acc do-prog iss-ess-ta. Pyengmin-ika Pyengmin-iuy kong-ul cha-ss-ta. past-decl Pyengmin-nom Pyengmin-gen ball-acc kick-past-decl Cengmin-ito Cengmin-iuy kong-ul cha-ss-ta. Cengmin-also Cengmin-gen ball-acc kick-past-decl ‘Pyengmin, Cengmin, and Tongmin were playing soccer in a playground. Pyeng- min kicked Pyengmin’s ball. Cengmin also kicked Cengmin’s ball.’ Pyengmin-ika ku-uy kong-ul cha-ss-ko, Cengmin-ito Pyengmin-nom he-gen ball-acc kick-past-conj Cengmin-also kuleha-yess-ta. so.do-past-decl ‘Pyengmin kicked his ball, and Cengmin did so, too.’

(10) Cinswu, Sungswu, Pyengswu-ka seythaksil-eyse seythak-ul ha-ko Cinswu Sungswu Pyengswu-nom laundry.room-at laundry-acc do-prog iss-ess-ta. Cinswu-ka Cinswu-uy seycey-lul sayongha-yss-ta. past-decl Cinswu-nom Cinswu-gen detergent-acc use-past-decl Sungswu-to Sungswu-uy seycey-lul sayongha-yss-ta. Sungswu-also Sungswu-gen detergent-acc use-past-decl ‘Cinswu, Sungswu, and Pyengswu were doing laundry at a laundry room. Cinswu used Cinswu’s detergent. Sungswu also used Sungswu’s detergent.’ Cinswu-ka ku-uy seycey-lul sayongha-yss-ko, Sungswu-to Cinswu-nom he-gen detergent-acc use-past-conj Sungswu-also kuleha-yess-ta. so.do-past-decl ‘Cinswu used his detergent, and Sungswu did so, too.’

(11) Kimkenmo, Kimcanghwun, Sinsunghwun-i nolaypang-eyse nol-ko Kimkenmo Kimcanghwun Sinsunghwun-nom singing.room-at hang.out-prog iss-ess-ta. Kimkenmo-ka Kimkenmo-uy hithukok-ul pwulu-ess-ta. past-decl Kimkenmo-nom Kimkenmo-gen hit.song-acc sing-past-decl Kimcanghwun-to Kimcanghwun-uy hithukok-ul pwulu-ess-ta. Kimcanghwun-also Kimcanghwun-gen hit.song-acc sing-past-decl ‘Kimkenmo, Kimcanghwun, and Sinsunghwun were hanging out in a singing room. Kimkenmo sang Kimkenmo’s hit song. Kimcanghwun also sang Kim- canghwun’s hit song.’ Kimkenmo-ka ku-uy hithukok-ul pwulu-ess-ko, Kimcanghwun-to Kimkenmo-nom he-gen hit.song-acc sing-past-conj Kimcanghwun-also kuleha-yess-ta. so.do-past-decl ‘Kimkenmo sang his hit song, and Kimcanghwun did so, too.’

180 (12) Pongswu, Thaykang, Thayho-ka pwuek-eyse selkeci-lul ha-ko Pongswu Thaykang Thayho-nom kitchen-at dish.washing-acc do-prog iss-ess-ta. Pongswu-ka Pongswu-uy khep-ul sis-ess-ta. past-decl Pongswu-nom Pongswu-gen cup-acc wash-past-decl Thaykang-ito Thaykang-iuy khep-ul sis-ess-ta. Thaykang-also Thaykang-gen cup-acc wash-past-decl ‘Pongswu, Thaykang, and Thayho were doing the dishes at a kitchen. Pongswu washed Pongswu’s cup. Thaykang also washed Thaykang’s cup.’ Pongswu-ka ku-uy khep-ul sis-ess-ko, Thaykang-ito Pongswu-nom he-gen cup-acc wash-past-conj Thaykang-also kuleha-yess-ta. so.do-past-decl ‘Pongswu washed his cup, and Thaykang did so, too.’

(13) Chiswu, Pyengswu, Hanswu-ka kkaphey-eyse tayhwua-lul nanwu-ko Chiswu Pyengswu Hanswu-nom café-at conversation-acc share-prog iss-ess-ta. Chiswu-ka Chiswu-uy cen yecachinkwu-lul past-decl Chiswu-nom Chiswu-gen previous girlfriend-acc helttut-ess-ta. Pyengswu-to Pyengswu-uy cen yecachinkwu-lul speak.ill.of-past-decl Pyengswu-also Pyengswu-gen previous girlfriend-acc helttut-ess-ta. speak.ill.of-past-decl ‘Chiswu, Pyengswu, and Hanswu were having a conversation at a café. Chiswu spoke ill of Chiswu’s ex-girlfriend. Pyengswu also spoke ill of Pyengswu’s ex- girlfriend.’ Chiswu-ka ku-uy cen yecachinkwu-lul helttut-ess-ko, Chiswu-nom he-gen previous girlfriend-acc speak.ill.of-past-conj Pyengswu-to kuleha-yess-ta. Pyengswu-also so.do-past-decl ‘Chiswu spoke ill of his ex-girlfriend, and Pyengswu did so, too.’

(14) Congswu, Hwuanswu, Tongswu-ka kongwuen-eyse nol-ko Congswu Hwuanswu Tongswu-nom park-at hang.out-prog iss-ess-ta. Congswu-ka Congswu-uy kay-lul kancilephi-ess-ta. past-decl Congswu-nom Congswu-gen dog-acc tickle-past-decl Hwuanswu-to Hwuanswu-uy kay-lul kancilephi-ess-ta. Hwuanswu-also Hwuanswu-gen dog-acc tickle-past-decl ‘Congswu, Hwuanswu, and Tongswu were hanging out at a park. Congswu tickled Congswu’s dog. Hwuanswu also tickled Hwuanswu’s dog.’ Congswu-ka ku-uy kay-lul kancilephi-ess-ko, Hwuanswu-to Congswu-nom he-gen dog-acc tickle-past-conj Hwuanswu-also kuleha-yess-ta. so.do-past-decl

181 ‘Congswu tickled his dog, and Hwuanswu did so, too.’

(15) Tongwu, Cinwu, Thaywu-ka Sinswu-uy cip-eyse nol-ko Tongwu Cinwu Thaywu-nom Sinswu-gen house-at hang.out-prog iss-ess-ta. Tongwu-ka Tongwu-uy yetongsayng-ul ttayli-ess-ta. past-decl Tongwu-nom Tongwu-gen younger.sister-acc hit-past-decl Cinwu-to Cinwu-uy yetongsayng-ul ttayli-ess-ta. Cinwu-also Cinwu-gen younger.sister-acc hit-past-decl ‘Tongwu, Cinwu, and Thaywu were hanging out at Sinswu’s house. Tongwu hit Tongwu’s younger sister. Cinwu also hit Cinwu’s younger sister.’ Tongwu-ka ku-uy yetongsayng-ul ttayli-ess-ko, Cinwu-to Tongwu-nom he-gen younger.sister-acc hit-past-conj Cinwu-also kuleha-yess-ta. so.do-past-decl ‘Tongwu hit his younger sister, and Cinwu did so, too.’

(16) Wonmwu, Kiswu, Sengswu-ka chatcip-eyse tayhwua-lul nanwu-ko Wonmwu Kiswu Sengswu-nom tea.house-at conversation-acc share-prog iss-ess-ta. Wonmwu-ka Wonmwu-uy nwuna-lul past-decl Wonmwu-nom Wonmwu-gen older.sister-acc chingchanha-yss-ta. Kiswu-to Kiswu-uy nwuna-lul compliment-past-decl Kiswu-also Kiswu-gen older.sister-acc chingchanha-yss-ta. compliment-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a tea house. Won- mwu complimented Wonmwu’s older sister. Kiswu also complimented Kiswu’s older sister.’ Wonmwu-ka ku-uy nwuna-lul chingchanha-yss-ko, Kiswu-to Wonmwu-nom he-gen older.sister-acc compliment-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl ‘Wonmwu complimented his older sister, and Kiswu did so, too.’

D.2 Condition 2: VPA-Free (strict identity reading)

182 Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to kuleha-yess-ta. Minswu-nom he-gen stuﬀ-acc move-past-conj Kiswu-also so.do-past-decl ‘Minswu moved his stuﬀ, and Kiswu did so, too.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse yakwu-lul ha-ko Yonghi Kenhi Myengho-nom play.ground-in baseball-acc do-prog iss-ess-ta. Yonghi-ka Yonghi-uy kong-ul tenci-ess-ta. Kenhi-to past-decl Yonghi-nom Yonghi-gen ball-acc throw-past-decl Kenhi-also Yonghi-uy kong-ul tenci-ess-ta. Yonghi-gen ball-acc throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground. Yonghi threw Yonghi’s ball. Kenhi also threw Yonghi’s ball.’ Yonghi-ka ku-uy kong-ul tenci-ess-ko, Kenhi-to kuleha-yess-ta. Yonghi-nom he-gen ball-acc throw-past-conj Kenhi-also so.do-past-decl ‘Yonghi threw his ball, and Kenhi did so, too.’

(3) Thayswu, Kiswu, Cengswu-ka pa-eyse wuain-ul masi-ko iss-ess-ta. Thayswu Kiswu Cengswu-nom bar-at wuain-acc drink-prog past-decl Thayswu-ka Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-to Thayswu-nom Thayswu-gen wine.glass-acc break-past-decl Kiswu-also Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Thayswu-gen wine.glass-acc break-past-decl ‘Thayswu, Kiswu, and Cengswu were drinking wine at a bar. Thayswu broke Thayswu’s wine glass. Kiswu also broke Thayswu’s wine glass.’ Thayswu-ka ku-uy wuaincan-ul kkaythuli-ess-ko, Kiswu-to Thayswu-nom he-gen wine.glass-acc break-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl ‘Thayswu broke his wine glass, and Kiswu did so, too.’

(4) Myengswu, Hyenswu, Cwunswu-ka wuntong hwu swui-ko iss-ess-ta. Myengswu Hyenswu Cwunswu-nom exercise after rest-prog past-decl Myengswu-ka Myengswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-to Myengswu-nom Myengswu-gen beverage-acc drink-past-decl Hyenswu-also Myengswu-uy umlyoswu-lul masi-ess-ta. Myengswu-gen beverage-acc drink-past-decl ‘Myengswu, Hyenswu, and Cwunswu were taking a rest after exercise. Myengswu drank Myengswu’s beverage. Hyenswu also drank Myengswu’s beverage.’ Myengswu-ka ku-uy umlyoswu-lul masi-ess-ko, Hyenswu-to Myengswu-nom he-gen beverage-acc drink-past-conj Hyenswu-also kuleha-yess-ta. so.do-past-decl

183 ‘Myengswu drank his beverage, and Hyenswu did so, too.’

(5) Pyengswu, Mwunswu, Cwunswu-ka cip-ul chengsoha-ko iss-ess-ta. Pyengswu Mwunswu Cwunswu-nom house-acc clean-prog past-decl Pyengswu-ka Pyengswu-uy pang-ul ssul-ess-ta. Mwunswu-to Pyengswu-nom Pyengswu-gen room-acc sweep-past-decl Mwunswu-also Pyengswu-uy pang-ul ssul-ess-ta. Pyengswu-gen room-acc sweep-past-decl ‘Pyengswu, Mwunswu, and Cwunswu were cleaning their house. Pyengswu swept Pyengswu’s room. Mwunswu also swept Pyengswu’s room.’ Pyengswu-ka ku-uy pang-ul ssul-ess-ko, Mwunswu-to Pyengswu-nom he-gen room-acc sweep-past-conj Mwunswu-also kuleha-yess-ta. so.do-past-decl ‘Pyengswu swept his room, and Mwunswu did so, too.’

(6) Chelswu, Yengswu, Cengswu-ka tosekwuan-eyse kongpwuha-ko iss-ess-ta. Chelswu Yengswu Cengswu-nom library-in study-prog past-decl Chelswu-ka Chelswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-to Chelswu-nom Chelswu-gen pen-acc drop-past-decl Yengswu-also Chelswu-uy pheyn-ul ttelethuli-ess-ta. Chelswu-gen pen-acc drop-past-decl ‘Chelswu, Yengswu, and Cengswu were studying in a library. Chelswu dropped Chelswu’s pen. Yengswu also dropped Chelswu’s pen.’ Chelswu-ka ku-uy pheyn-ul ttelethuli-ess-ko, Yengswu-to Chelswu-nom he-gen pen-acc drop-past-conj Chelswu-also kuleha-yess-ta. so.do-past-decl ‘Chelswu dropped his pen, and Yengswu did so, too.’

(7) Chelho, Phengho, Swuho-ka kakca phyenci-lul ssu-ko iss-ess-ta. Chelho Phengho Swuho-nom each letter-acc write-prog past-decl Chelho-ka Chelho-uy phyenci-lul ilk-ess-ta. Phengho-to Chelho-nom Chelho-gen letter-acc read-past-decl Phengho-also Chelho-uy phyenci-lul ilk-ess-ta. Chelho-gen letter-acc read-past-decl ‘Chelho, Phengho, and Swuho were each writing a letter. Chelho read Chelho’s letter. Phengho also read Chelho’s letter.’ Chelho-ka ku-uy phyenci-lul ilk-ess-ko, Phengho-to Chelho-nom he-gen letter-acc read-past-conj Phengho-also kuleha-yess-ta. so.do-past-decl

184 ‘Chelho read his letter, and Phengho did so, too.’

(8) Kwangho, Changho, Yengho-ka ilsiktang-eyse cemsim-ul mek-ko Kwangho Changho Yengho-nom Japanese.restaurant-at lunch-acc eat-prog iss-ess-ta. Kwangho-ka Kwangho-uy utong-ul mek-ess-ta. past-decl Kwangho-nom Kwangho-gen Udon-acc eat-past-decl Changho-to Kwangho-uy utong-ul mek-ess-ta. Changho-also Kwangho-gen Udon-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having lunch at a Japanese restaurant. Kwangho ate Kwangho’s Udon. Changho also ate Kwangho’s Udon.’ Kwangho-ka ku-uy utong-ul mek-ess-ko, Changho-to Kwangho-nom he-gen Udon-acc eat-past-conj Changho-also kuleha-yess-ta. so.do-past-decl ‘Kwangho ate his Udon, and Changho did so, too.’

(9) Pyengmin, Cengmin, Tongmin-ika wuntongcang-eyse chwukkwu-lul ha-ko Pyengmin Cengmin Tongmin-nom play.ground-in soccer-acc do-prog iss-ess-ta. Pyengmin-ika Pyengmin-iuy kong-ul cha-ss-ta. past-decl Pyengmin-nom Pyengmin-gen ball-acc kick-past-decl Cengmin-ito Pyengmin-iuy kong-ul cha-ss-ta. Cengmin-also Pyengmin-gen ball-acc kick-past-decl ‘Pyengmin, Cengmin, and Tongmin were playing soccer in a playground. Pyeng- min kicked Pyengmin’s ball. Cengmin also kicked Pyengmin’s ball.’ Pyengmin-ika ku-uy kong-ul cha-ss-ko, Cengmin-ito Pyengmin-nom he-gen ball-acc kick-past-conj Cengmin-also kuleha-yess-ta. so.do-past-decl ‘Pyengmin kicked his ball, and Cengmin did so, too.’

(10) Cinswu, Sungswu, Pyengswu-ka seythaksil-eyse seythak-ul ha-ko Cinswu Sungswu Pyengswu-nom laundry.room-at laundry-acc do-prog iss-ess-ta. Cinswu-ka Cinswu-uy seycey-lul sayongha-yss-ta. past-decl Cinswu-nom Cinswu-gen detergent-acc use-past-decl Sungswu-to Cinswu-uy seycey-lul sayongha-yss-ta. Sungswu-also Cinswu-gen detergent-acc use-past-decl ‘Cinswu, Sungswu, and Pyengswu were doing laundry at a laundry room. Cinswu used Cinswu’s detergent. Sungswu also used Cinswu’s detergent.’ Cinswu-ka ku-uy seycey-lul sayongha-yss-ko, Sungswu-to Cinswu-nom he-gen detergent-acc use-past-conj Sungswu-also kuleha-yess-ta. so.do-past-decl

185 ‘Cinswu used his detergent, and Sungswu did so, too.’

(11) Kimkenmo, Kimcanghwun, Sinsunghwun-i nolaypang-eyse nol-ko Kimkenmo Kimcanghwun Sinsunghwun-nom singing.room-at hang.out-prog iss-ess-ta. Kimkenmo-ka Kimkenmo-uy hithukok-ul pwulu-ess-ta. past-decl Kimkenmo-nom Kimkenmo-gen hit.song-acc sing-past-decl Kimcanghwun-to Kimkenmo-uy hithukok-ul pwulu-ess-ta. Kimcanghwun-also Kimkenmo-gen hit.song-acc sing-past-decl ‘Kimkenmo, Kimcanghwun, and Sinsunghwun were hanging out in a singing room. Kimkenmo sang Kimkenmo’s hit song. Kimcanghwun also sang Kimkenmo’s hit song.’ Kimkenmo-ka ku-uy hithukok-ul pwulu-ess-ko, Kimcanghwun-to Kimkenmo-nom he-gen hit.song-acc sing-past-conj Kimcanghwun-also kuleha-yess-ta. so.do-past-decl ‘Kimkenmo sang his hit song, and Kimcanghwun did so, too.’

(12) Pongswu, Thaykang, Thayho-ka pwuek-eyse selkeci-lul ha-ko Pongswu Thaykang Thayho-nom kitchen-at dish.washing-acc do-prog iss-ess-ta. Pongswu-ka Pongswu-uy khep-ul sis-ess-ta. past-decl Pongswu-nom Pongswu-gen cup-acc wash-past-decl Thaykang-ito Pongswu-uy khep-ul sis-ess-ta. Thaykang-also Pongswu-gen cup-acc wash-past-decl ‘Pongswu, Thaykang, and Thayho were doing the dishes at a kitchen. Pongswu washed Pongswu’s cup. Thaykang also washed Pongswu’s cup.’ Pongswu-ka ku-uy khep-ul sis-ess-ko, Thaykang-ito Pongswu-nom he-gen cup-acc wash-past-conj Thaykang-also kuleha-yess-ta. so.do-past-decl ‘Pongswu washed his cup, and Thaykang did so, too.’

186 Chiswu-ka ku-uy cen yecachinkwu-lul helttut-ess-ko, Chiswu-nom he-gen previous girlfriend-acc speak.ill.of-past-conj Pyengswu-to kuleha-yess-ta. Pyengswu-also so.do-past-decl ‘Chiswu spoke ill of his ex-girlfriend, and Pyengswu did so, too.’

(14) Congswu, Hwuanswu, Tongswu-ka kongwuen-eyse nol-ko Congswu Hwuanswu Tongswu-nom park-at hang.out-prog iss-ess-ta. Congswu-ka Congswu-uy kay-lul kancilephi-ess-ta. past-decl Congswu-nom Congswu-gen dog-acc tickle-past-decl Hwuanswu-to Congswu-uy kay-lul kancilephi-ess-ta. Hwuanswu-also Congswu-gen dog-acc tickle-past-decl ‘Congswu, Hwuanswu, and Tongswu were hanging out at a park. Congswu tickled Congswu’s dog. Hwuanswu also tickled Congswu’s dog.’ Congswu-ka ku-uy kay-lul kancilephi-ess-ko, Hwuanswu-to Congswu-nom he-gen dog-acc tickle-past-conj Hwuanswu-also kuleha-yess-ta. so.do-past-decl ‘Congswu tickled his dog, and Hwuanswu did so, too.’

(15) Tongwu, Cinwu, Thaywu-ka Sinswu-uy cip-eyse nol-ko Tongwu Cinwu Thaywu-nom Sinswu-gen house-at hang.out-prog iss-ess-ta. Tongwu-ka Tongwu-uy yetongsayng-ul ttayli-ess-ta. past-decl Tongwu-nom Tongwu-gen younger.sister-acc hit-past-decl Cinwu-to Tongwu-uy yetongsayng-ul ttayli-ess-ta. Cinwu-also Tongwu-gen younger.sister-acc hit-past-decl ‘Tongwu, Cinwu, and Thaywu were hanging out at Sinswu’s house. Tongwu hit Tongwu’s younger sister. Cinwu also hit Tongwu’s younger sister.’ Tongwu-ka ku-uy yetongsayng-ul ttayli-ess-ko, Cinwu-to Tongwu-nom he-gen younger.sister-acc hit-past-conj Cinwu-also kuleha-yess-ta. so.do-past-decl ‘Tongwu hit his younger sister, and Cinwu did so, too.’

187 ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a tea house. Won- mwu complimented Wonmwu’s older sister. Kiswu also complimented Wonmwu’s older sister.’ Wonmwu-ka ku-uy nwuna-lul chingchanha-yss-ko, Kiswu-to Wonmwu-nom he-gen older.sister-acc compliment-past-conj Kiswu-also kuleha-yess-ta. so.do-past-decl ‘Wonmwu complimented his older sister, and Kiswu did so, too.’

D.3 Condition 3: Quantiﬁcational-Bound

(1) Minswu, Kiswu, Cinswu-ka say kiswuksa-lo isaha-ko iss-ess-ta. Minswu Kiswu Cinswu-nom new dormitory-to move-prog past-decl Minswu-ka Minswu-uy cim-ul nalu-ess-ta. Kiswu-to Kiswu-uy Minswu-nom Minswu-gen stuff-acc move-past-decl Kiswu-also Kiswu-gen cim-ul nalu-ess-ta. Cinswu-to Cinswu-uy cim-ul stuff-acc move-past-decl Cinswu-also Cinswu-gen stuff-acc nalu-ess-ta. move-past-decl ‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuff. Kiswu also moved Kiswu’s stuff. Cinswu also moved Cinswu’s stuff.’ Motwu-ka ku-uy cim-ul nalu-ess-ta. everyone-nom he-gen stuff-acc move-past-decl ‘Everyone moved his stuff.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse yakwu-lul ha-ko Yonghi Kenhi Myengho-nom play.ground-in baseball-acc do-prog iss-ess-ta. Yonghi-ka Yonghi-uy kong-ul tenci-ess-ta. Kenhi-to past-decl Yonghi-nom Yonghi-gen ball-acc throw-past-decl Kenhi-also Kenhi-uy kong-ul tenci-ess-ta. Myengho-to Myengho-uy kong-ul Kenhi-gen ball-acc throw-past-decl Myengho-also Myengho-gen ball-acc tenci-ess-ta. throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground. Yonghi threw Yonghi’s ball. Kenhi also threw Kenhi’s ball. Myengho also threw Myengho’s ball.’ Motwu-ka ku-uy kong-ul tenci-ess-ta. everyone-nom he-gen ball-acc throw-past-decl ‘Everyone threw his ball.’

188 (3) Thayswu, Kiswu, Cengswu-ka pa-eyse wuain-ul masi-ko iss-ess-ta. Thayswu Kiswu Cengswu-nom bar-at wuain-acc drink-prog past-decl Thayswu-ka Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-to Thayswu-nom Thayswu-gen wine.glass-acc break-past-decl Kiswu-also Kiswu-uy wuaincan-ul kkaythuli-ess-ta. Cengswu-to Cengswu-uy Kiswu-gen wine.glass-acc break-past-decl Cengswu-also Cengswu-gen wuaincan-ul kkaythuli-ess-ta. wine.glass-acc break-past-decl ‘Thayswu, Kiswu, and Cengswu were drinking wine at a bar. Thayswu broke Thayswu’s wine glass. Kiswu also broke Kiswu’s wine glass. Cengswu also broke Cengswu’s wine glass.’ Motwu-ka ku-uy wuaincan-ul kkaythuli-ess-ta. everyone-nom he-gen wine.glass-acc break-past-decl ‘Everyone broke his wine glass.’

(4) Myengswu, Hyenswu, Cwunswu-ka wuntong hwu swui-ko iss-ess-ta. Myengswu Hyenswu Cwunswu-nom exercise after rest-prog past-decl Myengswu-ka Myengswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-to Myengswu-nom Myengswu-gen beverage-acc drink-past-decl Hyenswu-also Hyenswu-uy umlyoswu-lul masi-ess-ta. Cwunswu-to Cwunswu-uy Hyenswu-gen beverage-acc drink-past-decl Cwunswu-also Cwunswu-gen umlyoswu-lul masi-ess-ta. beverage-acc drink-past-decl ‘Myengswu, Hyenswu, and Cwunswu were taking a rest after exercise. Myengswu drank Myengswu’s beverage. Hyenswu also drank Hyenswu’s beverage. Cwunswu also drank Cwunswu’s beverage.’ Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’

(5) Pyengswu, Mwunswu, Cwunswu-ka cip-ul chengsoha-ko iss-ess-ta. Pyengswu Mwunswu Cwunswu-nom house-acc clean-prog past-decl Pyengswu-ka Pyengswu-uy pang-ul ssul-ess-ta. Mwunswu-to Pyengswu-nom Pyengswu-gen room-acc sweep-past-decl Mwunswu-also Mwunswu-uy pang-ul ssul-ess-ta. Cwunswu-to Cwunswu-uy Mwunswu-gen room-acc sweep-past-decl Cwunswu-also Cwunswu-gen pang-ul ssul-ess-ta. room-acc sweep-past-decl ‘Pyengswu, Mwunswu, and Cwunswu were cleaning their house. Pyengswu swept Pyengswu’s room. Mwunswu also swept Mwunswu’s room. Cwunswu also swept Cwunswu’s room.’ Motwu-ka ku-uy pang-ul ssul-ess-ta. everyone-nom he-gen room-acc sweep-past-decl

189 ‘Everyone swept his room.’

(6) Chelswu, Yengswu, Cengswu-ka tosekwuan-eyse kongpwuha-ko iss-ess-ta. Chelswu Yengswu Cengswu-nom library-in study-prog past-decl Chelswu-ka Chelswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-to Chelswu-nom Chelswu-gen pen-acc drop-past-decl Yengswu-also Yengswu-uy pheyn-ul ttelethuli-ess-ta. Cengswu-to Cengswu-uy pheyn-ul Yengswu-gen pen-acc drop-past-decl Cengswu-also Cengswu-gen pen-acc ttelethuli-ess-ta. drop-past-decl ‘Chelswu, Yengswu, and Cengswu were studying in a library. Chelswu dropped Chelswu’s pen. Yengswu also dropped Yengswu’s pen. Cengswu also dropped Cengswu’s pen.’ Motwu-ka ku-uy pheyn-ul ttelethuli-ess-ta. everyone-nom he-gen pen-acc drop-past-decl ‘Everyone dropped his pen.’

(7) Chelho, Phengho, Swuho-ka kakca phyenci-lul ssu-ko iss-ess-ta. Chelho Phengho Swuho-nom each letter-acc write-prog past-decl Chelho-ka Chelho-uy phyenci-lul ilk-ess-ta. Phengho-to Chelho-nom Chelho-gen letter-acc read-past-decl Phengho-also Phengho-uy phyenci-lul ilk-ess-ta. Swuho-to Swuho-uy phyenci-lul Phengho-gen letter-acc read-past-decl Swuho-also Swuho-gen letter-acc ilk-ess-ta. read-past-decl ‘Chelho, Phengho, and Swuho were each writing a letter. Chelho read Chelho’s letter. Phengho also read Phengho’s letter. Swuho also read Swuho’s letter.’ Motwu-ka ku-uy phyenci-lul ilk-ess-ta. everyone-nom he-gen letter-acc read-past-decl ‘Everyone read his letter.’

(8) Kwangho, Changho, Yengho-ka ilsiktang-eyse cemsim-ul mek-ko Kwangho Changho Yengho-nom Japanese.restaurant-at lunch-acc eat-prog iss-ess-ta. Kwangho-ka Kwangho-uy utong-ul mek-ess-ta. past-decl Kwangho-nom Kwangho-gen Udon-acc eat-past-decl Changho-to Changho-uy utong-ul mek-ess-ta. Yengho-to Changho-also Changho-gen Udon-acc eat-past-decl Yengho-also Yengho-uy utong-ul mek-ess-ta. Yengho-gen Udon-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having lunch at a Japanese restaurant. Kwangho ate Kwangho’s Udon. Changho also ate Changho’s Udon. Yengho also ate Yengho’s Udon.’ Motwu-ka ku-uy utong-ul mek-ess-ta. everyone-nom he-gen Udon-acc eat-past-decl

190 ‘Everyone ate his Udon.’

(9) Pyengmin, Cengmin, Tongmin-ika wuntongcang-eyse chwukkwu-lul ha-ko Pyengmin Cengmin Tongmin-nom play.ground-in soccer-acc do-prog iss-ess-ta. Pyengmin-ika Pyengmin-iuy kong-ul cha-ss-ta. past-decl Pyengmin-nom Pyengmin-gen ball-acc kick-past-decl Cengmin-ito Cengmin-iuy kong-ul cha-ss-ta. Tongmin-ito Cengmin-also Cengmin-gen ball-acc kick-past-decl Tongmin-also Tongmin-iuy kong-ul cha-ss-ta. Tongmin-gen ball-acc kick-past-decl ‘Pyengmin, Cengmin, and Tongmin were playing soccer in a playground. Pyeng- min kicked Pyengmin’s ball. Cengmin also kicked Cengmin’s ball. Tongmin also kicked Tongmin’s ball.’ Motwu-ka ku-uy kong-ul cha-ss-ta. everyone-nom he-gen ball-acc kick-past-decl ‘Everyone kicked his ball.’

(10) Cinswu, Sungswu, Pyengswu-ka seythaksil-eyse seythak-ul ha-ko Cinswu Sungswu Pyengswu-nom laundry.room-at laundry-acc do-prog iss-ess-ta. Cinswu-ka Cinswu-uy seycey-lul sayongha-yss-ta. past-decl Cinswu-nom Cinswu-gen detergent-acc use-past-decl Sungswu-to Sungswu-uy seycey-lul sayongha-yss-ta. Pyengswu-to Sungswu-also Sungswu-gen detergent-acc use-past-decl Pyengswu-also Pyengswu-uy seycey-lul sayongha-yss-ta. Pyengswu-gen detergent-acc use-past-decl ‘Cinswu, Sungswu, and Pyengswu were doing laundry at a laundry room. Cinswu used Cinswu’s detergent. Sungswu also used Sungswu’s detergent. Pyengswu also used Pyengswu’s detergent.’ Motwu-ka ku-uy seycey-lul sayongha-yss-ta. everyone-nom he-gen detergent-acc use-past-decl ‘Everyone used his detergent.’

(11) Kimkenmo, Kimcanghwun, Sinsunghwun-i nolaypang-eyse nol-ko Kimkenmo Kimcanghwun Sinsunghwun-nom singing.room-at hang.out-prog iss-ess-ta. Kimkenmo-ka Kimkenmo-uy hithukok-ul pwulu-ess-ta. past-decl Kimkenmo-nom Kimkenmo-gen hit.song-acc sing-past-decl Kimcanghwun-to Kimcanghwun-uy hithukok-ul pwulu-ess-ta. Kimcanghwun-also Kimcanghwun-gen hit.song-acc sing-past-decl Sinsunghwun-to Sinsunghwun-uy hithukok-ul pwulu-ess-ta. Sinsunghwun-also Sinsunghwun-gen hit.song-acc sing-past-decl ‘Kimkenmo, Kimcanghwun, and Sinsunghwun were hanging out in a singing room. Kimkenmo sang Kimkenmo’s hit song. Kimcanghwun also sang Kim- canghwun’s hit song. Sinsunghwun also sang Sinsunghwun’s hit song.’

191 Motwu-ka ku-uy hithukok-ul pwulu-ess-ta. everyone-nom he-gen hit.song-acc sing-past-decl ‘Everyone sang his hit song.’

(12) Pongswu, Thaykang, Thayho-ka pwuek-eyse selkeci-lul ha-ko Pongswu Thaykang Thayho-nom kitchen-at dish.washing-acc do-prog iss-ess-ta. Pongswu-ka Pongswu-uy khep-ul sis-ess-ta. past-decl Pongswu-nom Pongswu-gen cup-acc wash-past-decl Thaykang-ito Thaykang-iuy khep-ul sis-ess-ta. Thayho-to Thaykang-also Thaykang-gen cup-acc wash-past-decl Thayho-also Thayho-uy khep-ul sis-ess-ta. Thayho-gen cup-acc wash-past-decl ‘Pongswu, Thaykang, and Thayho were doing the dishes at a kitchen. Pongswu washed Pongswu’s cup. Thaykang also washed Thaykang’s cup. Thayho also washed Thayho’s cup.’ Motwu-ka ku-uy khep-ul sis-ess-ta. everyone-nom he-gen cup-acc wash-past-decl ‘Everyone washed his cup.’

The four test items that are not given here are the same as those for the Bound-September condition in Experiment 2.

D.4 Condition 4: Quantiﬁcational-Free

(1) Minswu, Kiswu, Cinswu-ka Thayswu-ka oki-lul kitalimye, say Minswu Kiswu Cinswu-nom Thayswu-nom coming-acc waiting new kiswuksa-lo isaha-ko iss-ess-ta. Minswu-ka Thayswu-uy cim-ul dormitory-to move-prog past-decl Minswu-nom Thayswu-gen stuff-acc nalu-ess-ta. Kiswu-to Thayswu-uy cim-ul nalu-ess-ta. move-past-decl Kiswu-also Thayswu-gen stuff-acc move-past-decl Cinswu-to Thayswu-uy cim-ul nalu-ess-ta. Cinswu-also Thayswu-gen stuff-acc move-past-decl ‘Minswu, Kiswu, and Cinswu were moving to a new dormitory, waiting for Thayswu to come. Minswu moved Thayswu’s stuff. Kiswu also moved Thayswu’s stuff. Cinswu also moved Thayswu’s stuff.’ Motwu-ka ku-uy cim-ul nalu-ess-ta. everyone-nom he-gen stuff-acc move-past-decl ‘Everyone moved his stuff.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse, Cinswu-ka oki-lul Yonghi Kenhi Myengho-nom play.ground-in Cinswu-nom coming-acc kitalimye, yakwu-lul ha-ko iss-ess-ta. Yonghi-ka Cinswu-uy waiting baseball-acc do-prog past-decl Yonghi-nom Cinswu-gen

192 kong-ul tenci-ess-ta. Kenhi-to Cinswu-uy kong-ul tenci-ess-ta. ball-acc throw-past-decl Kenhi-also Cinswu-gen ball-acc throw-past-decl Myengho-to Cinswu-uy kong-ul tenci-ess-ta. Myengho-also Cinswu-gen ball-acc throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground, waiting for Cinswu to come. Yonghi threw Cinswu’s ball. Kenhi also threw Cinswu’s ball. Myengho also threw Cinswu’s ball.’ Motwu-ka ku-uy kong-ul tenci-ess-ta. everyone-nom he-gen ball-acc throw-past-decl ‘Everyone threw his ball.’

(3) Thayswu, Kiswu, Cengswu-ka pa-eyse, Yonghi-ka oki-lul kitalimye, Thayswu Kiswu Cengswu-nom bar-at Yonghi-nom coming-acc waiting wuain-ul masi-ko iss-ess-ta. Thayswu-ka Yonghi-uy wuaincan-ul wuain-acc drink-prog past-decl Thayswu-nom Yonghi-gen wine.glass-acc kkaythuli-ess-ta. Kiswu-to Yonghi-uy wuaincan-ul kkaythuli-ess-ta. break-past-decl Kiswu-also Yonghi-gen wine.glass-acc break-past-decl Cengswu-to Yonghi-uy wuaincan-ul kkaythuli-ess-ta. Cengswu-also Yonghi-gen wine.glass-acc break-past-decl ‘Thayswu, Kiswu, and Cengswu were drinking wine at a bar, waiting for Yonghi to come. Thayswu broke Yonghi’s wine glass. Kiswu also broke Yonghi’s wine glass. Cengswu also broke Yonghi’s wine glass.’ Motwu-ka ku-uy wuaincan-ul kkaythuli-ess-ta. everyone-nom he-gen wine.glass-acc break-past-decl ‘Everyone broke his wine glass.’

(4) Myengswu, Hyenswu, Cwunswu-ka wuntong hwu, Kiswu-ka oki-lul Myengswu Hyenswu Cwunswu-nom exercise after Kiswu-nom coming-acc kitalimye, swui-ko iss-ess-ta. Myengswu-ka Kiswu-uy umlyoswu-lul waiting rest-prog past-decl Myengswu-nom Kiswu-gen beverage-acc masi-ess-ta. Hyenswu-to Kiswu-uy umlyoswu-lul masi-ess-ta. drink-past-decl Hyenswu-also Kiswu-gen beverage-acc drink-past-decl Cwunswu-to Kiswu-uy umlyoswu-lul masi-ess-ta. Cwunswu-also Kiswu-gen beverage-acc drink-past-decl ‘Myengswu, Hyenswu, and Cwunswu were taking a rest after exercise, waiting for Kiswu to come. Myengswu drank Kiswu’s beverage. Hyenswu also drank Kiswu’s beverage. Cwunswu also drank Kiswu’s beverage.’ Motwu-ka ku-uy umlyoswu-lul masi-ess-ta. everyone-nom he-gen beverage-acc drink-past-decl ‘Everyone drank his beverage.’

193 (5) Pyengswu, Mwunswu, Cwunswu-ka, Sengswu-ka oki-lul kitalimye, Pyengswu Mwunswu Cwunswu-nom Sengswu-nom coming-acc waiting cip-ul chengsoha-ko iss-ess-ta. Pyengswu-ka Sengswu-uy pang-ul house-acc clean-prog past-decl Pyengswu-nom Sengswu-gen room-acc ssul-ess-ta. Mwunswu-to Sengswu-uy pang-ul ssul-ess-ta. sweep-past-decl Mwunswu-also Sengswu-gen room-acc sweep-past-decl Cwunswu-to Sengswu-uy pang-ul ssul-ess-ta. Cwunswu-also Sengswu-gen room-acc sweep-past-decl ‘Pyengswu, Mwunswu, and Cwunswu were cleaning the house, waiting for Sen- gswu to come. Pyengswu swept Sengswu’s room. Mwunswu also swept Sengswu’s room. Cwunswu also swept Sengswu’s room.’ Motwu-ka ku-uy pang-ul ssul-ess-ta. everyone-nom he-gen room-acc sweep-past-decl ‘Everyone swept his room.’

(6) Chelswu, Yengswu, Cengswu-ka tosekwuan-eyse, Pemswu-ka oki-lul Chelswu Yengswu Cengswu-nom library-in Pemswu-nom coming-acc kitalimye, kongpwuha-ko iss-ess-ta. Chelswu-ka Pemswu-uy pheyn-ul waiting study-prog past-decl Chelswu-nom Pemswu-gen pen-acc ttelethuli-ess-ta. Yengswu-to Pemswu-uy pheyn-ul ttelethuli-ess-ta. drop-past-decl Yengswu-also Pemswu-gen pen-acc drop-past-decl Cengswu-to Pemswu-uy pheyn-ul ttelethuli-ess-ta. Cengswu-also Pemswu-gen pen-acc drop-past-decl ‘Chelswu, Yengswu, and Cengswu were studying in a library, waiting for Pemswu to come. Chelswu dropped Pemswu’s pen. Yengswu also dropped Pemswu’s pen. Cengswu also dropped Pemswu’s pen.’ Motwu-ka ku-uy pheyn-ul ttelethuli-ess-ta. everyone-nom he-gen pen-acc drop-past-decl ‘Everyone dropped his pen.’

(7) Chelho, Phengho, Swuho-ka, Cengswu-ka oki-lul kitalimye, kakca Chelho Phengho Swuho-nom Cengswu-nom coming-acc waiting each phyenci-lul ssu-ko iss-ess-ta. Chelho-ka Cengswu-uy phyenci-lul letter-acc write-prog past-decl Chelho-nom Cengswu-gen letter-acc ilk-ess-ta. Phengho-to Cengswu-uy phyenci-lul ilk-ess-ta. read-past-decl Phengho-also Cengswu-gen letter-acc read-past-decl Swuho-to Cengswu-uy phyenci-lul ilk-ess-ta. Swuho-also Cengswu-gen letter-acc read-past-decl ‘Chelho, Phengho, and Swuho were each writing a letter, waiting for Cengswu to come. Chelho read Cengswu’s letter. Phengho also read Cengswu’s letter. Swuho also read Cengswu’s letter.’ Motwu-ka ku-uy phyenci-lul ilk-ess-ta. everyone-nom he-gen letter-acc read-past-decl

194 ‘Everyone read his letter.’

(8) Kwangho, Changho, Yengho-ka ilsiktang-eyse, Sangki-ka Kwangho Changho Yengho-nom Japanese.restaurant-at Sangki-nom oki-lul kitalimye, cemsim-ul mek-ko iss-ess-ta. Kwangho-ka coming-acc waiting lunch-acc eat-prog past-decl Kwangho-nom Sangki-uy utong-ul mek-ess-ta. Changho-to Sangki-uy utong-ul Sangki-gen Udon-acc eat-past-decl Changho-also Sangki-gen Udon-acc mek-ess-ta. Yengho-to Sangki-uy utong-ul mek-ess-ta. eat-past-decl Yengho-also Sangki-gen Udon-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having lunch at a Japanese restaurant, waiting for Sangki to come. Kwangho ate Sangki’s Udon. Changho also ate Sangki’s Udon. Yengho also ate Sangki’s Udon.’ Motwu-ka ku-uy utong-ul mek-ess-ta. everyone-nom he-gen Udon-acc eat-past-decl ‘Everyone ate his Udon.’

The eight test items that are not given here are the same as those for the Free-August and Free-September conditions in Experiment 2.

195 Appendix E

Test sentences from Experiment 4

E.1 Condition 1: Simple-Extraction (short distance scrambling)

(1) Minswu-wa Hanswu-ka kakca etten kwamok-ul yelsimhi Minswu-and Hanswu-nom each which subject-acc hard kongpwuha-yess-nunci malha-ko iss-ess-ta. study-past-comp talk-prog past-decl ‘Minswu and Hanswu were talking about which subject they each studied hard.’

M: Swuhak-ul1 na-nun [VP yelsimhi t1 kongpwuha]-yss-e. Ne-nun? math-acc I-top hard study-past-decl you-top ‘I studied math hard. How about you?’ H: Swuhak-ul na-to kulay-ss-e. math-acc I-also so.do-past-decl Intended: ‘I studied math hard, too.’

(2) Pongswu-wa Hyentong-ika kakca etten kkoch-ul yelsimhi Pongswu-and Hyentong-nom each which ﬂower-acc diligently sim-ess-nunci malha-ko iss-ess-ta. plant-past-comp talk-prog past-decl ‘Pongswu and Hyentong were talking about which ﬂowers they each planted diligently.’

P: Haypalaki-lul1 na-nun [VP yelsimhi t1 sim]-ess-e. Ne-nun? sunflower-acc I-top diligently plant-past-decl you-top ‘I planted sunflowers diligently. How about you?’ H: Haypalaki-lul na-to kulay-ss-e. sunflower-acc I-also so.do-past-decl Intended: ‘I planted sunflowers diligently, too.’

196 (3) Sungho-wa Changcin-ika kakca etten cencaceyphwum-ul olay Sungho-and Changcin-nom each which electronic.goods-acc long sayongha-yess-nunci malha-ko iss-ess-ta. use-past-comp talk-prog past-decl ‘Sungho and Changcin were talking about which electronic goods they each used for a long period of time.’

S: Nothupwuk-ul1 na-nun [VP olay t1 sayongha]-yss-e. Ne-nun? laptop-acc I-top long use-past-decl you-top ‘I used a laptop for a long period of time. How about you?’ C: Nothupwuk-ul na-to kulay-ss-e. laptop-acc I-also so.do-past-decl Intended: ‘I used a laptop for a long period of time, too.’

(4) Kyuho-wa Sangchel-ika kakca etten mwuswul-ul olay Kyuho-and Sangchel-nom each which martial.arts-acc long swulyenha-yess-nunci malha-ko iss-ess-ta. practice-past-comp talk-prog past-decl ‘Kyuho and Sangchel were talking about which martial arts they each practiced for a long period of time.’

K: Thaykwuento-lul1 na-nun [VP olay t1 swulyenha]-yss-e. Ne-nun? Taekwondo-acc I-top long practice-past-decl you-top ‘I practiced Taekwondo for a long period of time. How about you?’ S: Thaykwuento-lul na-to kulay-ss-e. Taekwondo-acc I-also so.do-past-decl Intended: ‘I practiced Taekwondo for a long period of time, too.’

(5) Kyengphyo-wa Caychel-ika kakca etten sociphwum-ul cacwu Kyengphyo-and Caychel-nom each which belongings-acc frequently ilepeli-ess-nunci malha-ko iss-ess-ta. lose-past-comp talk-prog past-decl ‘Kyengphyo and Caychel were talking about which belongings they each lost frequently.’

K: Cikap-ul1 na-nun [VP cacwu t1 ilepeli]-ess-e. Ne-nun? wallet-acc I-top frequently lose-past-decl you-top ‘I lost a wallet freqeuntly. How about you?’ C: Cikap-ul na-to kulay-ss-e. wallet-acc I-also so.do-past-decl Intended: ‘I lost a wallet frequently, too.’

197 (6) Congswu-wa Saypem-ika kakca etten cha-ul cengmal coaha-yess-nunci Congswu-and Saypem-nom each which car-acc really like-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Congswu and Saypem were talking about which car they each really liked.’

C: Nywupithul-ul1 na-nun [VP cengmal t1 coaha]-yss-e. Ne-nun? New.Beetle-acc I-top really like-past-decl you-top ‘I really liked New Beetle. How about you?’ S: Nywupithul-ul na-to kulay-ss-e. New.Beetle-acc I-also so.do-past-decl Intended: ‘I really liked New Beetle, too.’

(7) Swuho-wa Cayhak-ika kakca etten panchan-ul cengmal Swuho-and Cayhak-nom each which side.dish-acc really sileha-yess-nunci malha-ko iss-ess-ta. dislike-past-comp talk-prog past-decl ‘Swuho and Cayhak were talking about which side dish they each really disliked.’

S: Kimchi-lul1 na-nun [VP cengmal t1 sileha]-yss-e. Ne-nun? Kimchi-acc I-top really dislike-past-decl you-top ‘I really disliked Kimchi. How about you?’ C: Kimchi-lul na-to kulay-ss-e. Kimchi-acc I-also so.do-past-decl Intended: ‘I really disliked Kimchi, too.’

(8) Congwu-wa Taykwan-ika kakca etten homsyophing sangphwum-ul Congwu-and Taykwan-nom each which home.shopping goods-acc cacwu cwumwunha-yess-nunci malha-ko iss-ess-ta. frequently order-past-comp talk-prog past-decl ‘Congwu and Taykwan were talking about which home shopping goods they each ordered frequently.’

C: Kimchi-lul1 na-nun [VP cacwu t1 cwumwunha]-yss-e. Ne-nun? Kimchi-acc I-top frequently order-past-decl you-top ‘I ordered Kimchi frequently. How about you?’ T: Kimchi-lul na-to kulay-ss-e. Kimchi-acc I-also so.do-past-decl Intended: ‘I ordered Kimchi frequently, too.’

198 (9) Bemswu-wa Twusik-ika kakca etten thipiphulo-lul cacwu Bemswu-and Twusik-nom each which television.program-acc frequently sichengha-yess-nunci malha-ko iss-ess-ta. watch-past-comp talk-prog past-decl ‘Bemswu and Twusik were talking about which TV show they each watched frequently.’

B: Mwuhantocen-ul1 na-nun [VP cacwu t1 sichengha]-yss-e. Infinite.Challenge-acc I-top frequently watch-past-decl Ne-nun? you-top ‘I watched Infinite Challenge frequently. How about you?’ T: Mwuhantocen-ul na-to kulay-ss-e. Infinite.Challenge-acc I-also so.do-past-decl Intended: ‘I watched Infinite Challenge frequently, too.’

(10) Chelswu-wa Heechel-ika kakca etten maykcwu-lul cacwu Chelswu-and Heechel-nom each which beer-acc frequently masi-ess-nunci malha-ko iss-ess-ta. drink-past-comp talk-prog past-decl ‘Chelswu and Heechel were talking about which beer they each drank frequently.’

C: Chasu-lul1 na-nun [VP cacwu t1 masi]-ess-e. Ne-nun? Cass-acc I-top frequently drink-past-decl you-top ‘I drank Cass frequently. How about you?’ H: Chasu-lul na-to kulay-ss-e. Cass-acc I-also so.do-past-decl Intended: ‘I drank Cass frequently, too.’

(11) Minswu-wa Yengmin-ika kakca etten ilponumsik-ul cacwu Minswu-and Yengmin-nom each which Japanese.food-acc frequently mek-ess-nunci malha-ko iss-ess-ta. eat-past-comp talk-prog past-decl ‘Minswu and Yengmin were talking about which Japanese food they each ate frequently.’

M: Udong-ul1 na-nun [VP cacwu t1 mek]-ess-e. Ne-nun? Udon-acc I-top frequently eat-past-decl you-top ‘I ate Udon frequently. How about you?’ Y: Udong-ul na-to kulay-ss-e. Udon-acc I-also so.do-past-decl Intended: ‘I ate Udon frequently, too.’

199 (12) Sengmo-wa Cinchel-ika kakca etten sinmwun-ul olay Sengmo-and Cinchel-nom each which newspaper-acc long kwutokha-yess-nunci malha-ko iss-ess-ta. subscribe-past-comp talk-prog past-decl ‘Sengmo and Cinchel were talking about which newspaper they each subscribed to for a long period of time.’

S: Tongailpo-lul1 na-nun [VP olay t1 kwutokha]-yss-e. Ne-nun? Dong-A.Ilbo-acc I-top long subscribe-past-decl you-top ‘I subscribed to Dong-A Ilbo for a long period of time. How about you?’ C: Tongailpo-lul na-to kulay-ss-e. DongA.Ilbo-acc I-also so.do-past-decl Intended: ‘I subscribed to Dong-A Ilbo for a long period of time, too.’

(13) Hyenho-wa Tayhwun-ika kakca etten yenghwa-lul cacwu Hyenho-and Tayhwun-nom each which movie-acc frequently poa-ss-nunci malha-ko iss-ess-ta. watch-past-comp talk-prog past-decl ‘Hyenho and Tayhwun were talking about which movie they each watched frequently.’

H: Phulocun-ul1 na-nun [VP cacwu t1 poa]-ss-e. Ne-nun? Frozen-acc I-top frequently watch-past-decl you-top ‘I watched Frozen frequently. How about you?’ T: Phulocun-ul na-to kulay-ss-e. Frozen-acc I-also so.do-past-decl Intended: ‘I watched Frozen frequently, too.’

(14) Thayswu-wa Myenghwun-ika kakca etten akki-lul olay Thayswu-and Myenghwun-nom each which music.instrument-acc long yencwuha-yess-nunci malha-ko iss-ess-ta. play-past-comp talk-prog past-decl ‘Thayswu and Myenghwun were talking about which music instrument they each played for a long period of time.’

T: Phiano-lul1 na-nun [VP olay t1 yencwuha]-yss-e. Ne-nun? piano-acc I-top long play-past-decl you-top ‘I played the piano for a long period of time. How about you?’ M: Phiano-lul na-to kulay-ss-e. piano-acc I-also so.do-past-decl Intended: ‘I played the piano for a long period of time, too.’

200 (15) Hungswu-wa Minhyen-ika kakca etten hankwukumsik-ul cacwu Hungswu-and Minhyen-nom each which Korean.food-acc frequently yoliha-yess-nunci malha-ko iss-ess-ta. cook-past-comp talk-prog past-decl ‘Hungswu and Minhyen were talking about which Korean food they each cooked frequently.’

H: Kimchi-lul1 na-nun [VP cacwu t1 yoliha]-yss-e. Ne-nun? Kimchi-acc I-top frequently cook-past-decl you-top ‘I cooked Kimchi frequently. How about you?’ M: Kimchi-lul na-to kulay-ss-e. Kimchi-acc I-also so.do-past-decl Intended: ‘I cooked Kimchi frequently, too.’

(16) Unphyo-wa Congsin-ika kakca etten nolay-lul cacwu pwulu-ess-nunci Unphyo-and Congsin-nom each which song-acc frequently sing-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Unphyo and Congsin were talking about which song they each sang frequently.’

U: Mannam-ul1 na-nun [VP cacwu t1 pwulu]-ess-e. Ne-nun? Mannam-acc I-top frequently sing-past-decl you-top ‘I sang Mannam frequently. How about you?’ C: Mannam-ul na-to kulay-ss-e. Mannam-acc I-also so.do-past-decl Intended: ‘I sang Mannam frequently, too.’

(17) Kithay-wa Cenghwun-ika kakca etten nolay-lul cacwu Kithay-and Cenghwun-nom each which song-acc frequently tul-ess-nunci malha-ko iss-ess-ta. listen-past-comp talk-prog past-decl ‘Kithay and Cenghwun were talking about which song they each listened to frequently.’

K: Mannam-ul1 na-nun [VP cacwu t1 tul]-ess-e. Ne-nun? Mannam-acc I-top frequently listen-past-decl you-top ‘I listened to Mannam frequently. How about you?’ C: Mannam-ul na-to kulay-ss-e. Mannam-acc I-also so.do-past-decl Intended: ‘I listened to Mannam frequently, too.’

201 (18) Kyengho-wa Hakpem-ika kakca etten kwail-ul cacwu sa-ss-nunci Kyengho-and Hakpem-nom each which fruit-acc frequently buy-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Kyengho and Hakpem were talking about which fruit they each bought frequently.’

K: Sakwa-lul1 na-nun [VP cacwu t1 sa]-ss-e. Ne-nun? apple-acc I-top frequently buy-past-decl you-top ‘I bought apples frequently. How about you?’ H: Sakwa-lul na-to kulay-ss-e. apple-acc I-also so.do-past-decl Intended: ‘I bought apples frequently, too.’

(19) Hyenwu-wa Tonghwun-ika kakca etten chwum-ul mayil Hyenwu-and Tonghwun-nom each which dance-acc every.day chwu-ess-nunci malha-ko iss-ess-ta. dance-past-comp talk-prog past-decl ‘Hyenwu and Tonghwun were talking about which dance they each danced everyday.’

H: Thayngko-lul1 na-nun [VP mayil t1 chwu]-ess-e. Ne-nun? Tango-acc I-top every.day dance-past-decl you-top ‘I danced Tango everyday. How about you?’ T: Thayngko-lul na-to kulay-ss-e. Tango-acc I-also so.do-past-decl Intended: ‘I danced Tango everyday, too.’

(20) Ciseng-iwa Mincong-ika kakca etten aywantongmwul-ul olay Ciseng-and Mincong-nom each which pet-acc long khiwu-ess-nunci malha-ko iss-ess-ta. raise-past-comp talk-prog past-decl ‘Ciseng and Mincong were talking about which pet they each kept for a long period of time.’

C: Kay-lul1 na-nun [VP olay t1 khiwu]-ess-e. Ne-nun? dog-acc I-top long raise-past-decl you-top ‘I kept a dog for a long period of time. How about you?’ M: Kay-lul na-to kulay-ss-e. dog-acc I-also so.do-past-decl Intended: ‘I kept a dog for a long period of time, too.’

202 (21) Henswu-wa Canghwun-ika kakca etten mwulken-ul olay mo-ass-nunci Henswu-and Canghwun-nom each which item-acc long collect-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Henswu and Canghwun were talking about which item they each collected for a long period of time.’

H: Wuphyo-lul1 na-nun [VP olay t1 mo]-ass-e. Ne-nun? stamp-acc I-top long collect-past-decl you-top ‘I collected stamps for a long period of time. How about you?’ C: Wuphyo-lul na-to kulay-ss-e. stamp-acc I-also so.do-past-decl Intended: ‘I collected stamps for a long period of time, too.’

(22) Changho-wa Minhyen-ika kakca etten wuekuke-lul olay Changho-and Minhyen-nom each which foreign.language-acc long paywu-ess-nunci malha-ko iss-ess-ta. learn-past-comp talk-prog past-decl ‘Changho and Minhyen were talking about which foreign language they each learned for a long period of time.’

C: Pwule-lul1 na-nun [VP olay t1 paywu]-ess-e. Ne-nun? French-acc I-top long learn-past-decl you-top ‘I learned French for a long period of time. How about you?’ M: Pwule-lul na-to kulay-ss-e. French-acc I-also so.do-past-decl Intended: ‘I learned French for a long period of time, too.’

(23) Cengho-wa Senghyen-ika kakca etten ppang-ul cacwu Cengho-and Senghyen-nom each which bread-acc frequently kwu-ess-nunci malha-ko iss-ess-ta. bake-past-comp talk-prog past-decl ‘Cengho and Senghyen were talking about which bread they each baked frequently.’

C: Manulppang-ul1 na-nun [VP cacwu t1 kwu]-ess-e. Ne-nun? garlic.bread-acc I-top frequently bake-past-decl you-top ‘I baked garlic bread frequently. How about you?’ S: Manulppang-ul na-to kulay-ss-e. garlic.bread-acc I-also so.do-past-decl Intended: ‘I baked garlic bread frequently, too.’

203 (24) Tonghay-wa Siwan-ika kakca etten chayk-ul cacwu ilk-ess-nunci Tonghay-and Siwan-nom each which book-acc frequently read-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Tonghay and Siwan were talking about which book they each read frequently.’

T: Hayliphothe-lul1 na-nun [VP cacwu t1 ilk]-ess-e. Ne-nun? Harry.Potter-acc I-top frequently read-past-decl you-top ‘I read Harry Potter frequently. How about you?’ S: Hayliphothe-lul na-to kulay-ss-e. Harry.Potter-acc I-also so.do-past-decl Intended: ‘I read Harry Potter frequently, too.’

E.2 Condition 2: Simple-NoExtraction

M: Na-nun [VP yelsimhi swuhak-ul kongpwuha]-yss-e. Ne-nun? I-top hard math-acc study-past-decl you-top ‘I studied math hard. How about you?’ H: Na-to kulay-ss-e. I-also do.so-past-decl Intended: ‘I studied math hard, too.’

P: Na-nun [VP yelsimhi haypalaki-lul sim]-ess-e. Ne-nun? I-top diligently sunflower-acc plant-past-decl you-top ‘I planted sunflowers diligently. How about you?’ H: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I planted sunflowers diligently, too.’

204 (3) Sungho-wa Changcin-ika kakca etten cencaceyphwum-ul olay Sungho-and Changcin-nom each which electronic.goods-acc long sayongha-yess-nunci malha-ko iss-ess-ta. use-past-comp talk-prog past-decl ‘Sungho and Changcin were talking about which electronic goods they each used for a long period of time.’

S: Na-nun [VP olay nothupwuk-ul sayongha]-yss-e. Ne-nun? I-top long laptop-acc use-past-decl you-top ‘I used a laptop for a long period of time. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I used a laptop for a long period of time, too.’

K: Na-nun [VP olay thaykwuento-lul swulyenha]-yss-e. Ne-nun? I-top long Taekwondo-acc practice-past-decl you-top ‘I practiced Taekwondo for a long period of time. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I practiced Taekwondo for a long period of time, too.’

K: Na-nun [VP cacwu cikap-ul ilepeli]-ess-e. Ne-nun? I-top frequently wallet-acc lose-past-decl you-top ‘I lost a wallet freqeuntly. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I lost a wallet frequently, too.’

205 (6) Congswu-wa Saypem-ika kakca etten cha-ul cengmal coaha-yess-nunci Congswu-and Saypem-nom each which car-acc really like-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Congswu and Saypem were talking about which car they each really liked.’

C: Na-nun [VP cengmal nywupithul-ul coaha]-yss-e. Ne-nun? I-top really New.Beetle-acc like-past-decl you-top ‘I really liked New Beetle. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I really liked New Beetle, too.’

S: Na-nun [VP cengmal kimchi-lul sileha]-yss-e. Ne-nun? I-top really Kimchi-acc dislike-past-decl you-top ‘I really disliked Kimchi. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I really disliked Kimchi, too.’

C: Na-nun [VP cacwu kimchi-lul cwumwunha]-yss-e. Ne-nun? I-top frequently Kimchi-acc order-past-decl you-top ‘I ordered Kimchi frequently. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I ordered Kimchi frequently, too.’

206 (9) Bemswu-wa Twusik-ika kakca etten thipiphulo-lul cacwu Bemswu-and Twusik-nom each which television.program-acc frequently sichengha-yess-nunci malha-ko iss-ess-ta. watch-past-comp talk-prog past-decl ‘Bemswu and Twusik were talking about which TV show they each watched frequently.’

B: Na-nun [VP cacwu mwuhantocen-ul sichengha]-yss-e. I-top frequently Infinite.Challenge-acc watch-past-decl Ne-nun? you-top ‘I watched Infinite Challenge frequently. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I watched Infinite Challenge frequently, too.’

C: Na-nun [VP cacwu chasu-lul masi]-ess-e. Ne-nun? I-top frequently Cass-acc drink-past-decl you-top ‘I drank Cass frequently. How about you?’ H: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I drank Cass frequently, too.’

M: Na-nun [VP cacwu udong-ul mek]-ess-e. Ne-nun? I-top frequently Udon-acc eat-past-decl you-top ‘I ate Udon frequently. How about you?’ Y: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I ate Udon frequently, too.’

207 (12) Sengmo-wa Cinchel-ika kakca etten sinmwun-ul olay Sengmo-and Cinchel-nom each which newspaper-acc long kwutokha-yess-nunci malha-ko iss-ess-ta. subscribe-past-comp talk-prog past-decl ‘Sengmo and Cinchel were talking about which newspaper they each subscribed to for a long period of time.’

S: Na-nun [VP olay tongailpo-lul kwutokha]-yss-e. Ne-nun? I-top long Dong-A.Ilbo-acc subscribe-past-decl you-top ‘I subscribed to Dong-A Ilbo for a long period of time. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I subscribed to Dong-A Ilbo for a long period of time, too.’

H: Na-nun [VP cacwu phulocun-ul poa]-ss-e. Ne-nun? I-top frequently Frozen-acc watch-past-decl you-top ‘I watched Frozen frequently. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I watched Frozen frequently, too.’

T: Na-nun [VP olay phiano-lul yencwuha]-yss-e. Ne-nun? I-top long piano-acc play-past-decl you-top ‘I played the piano for a long period of time. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I played the piano for a long period of time, too.’

208 (15) Hungswu-wa Minhyen-ika kakca etten hankwukumsik-ul cacwu Hungswu-and Minhyen-nom each which Korean.food-acc frequently yoliha-yess-nunci malha-ko iss-ess-ta. cook-past-comp talk-prog past-decl ‘Hungswu and Minhyen were talking about which Korean food they each cooked frequently.’

H: Na-nun [VP cacwu kimchi-lul yoliha]-yss-e. Ne-nun? I-top frequently Kimchi-acc cook-past-decl you-top ‘I cooked Kimchi frequently. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I cooked Kimchi frequently, too.’

U: Na-nun [VP cacwu mannam-ul pwulu]-ess-e. Ne-nun? I-top frequently Mannam-acc sing-past-decl you-top ‘I sang Mannam frequently. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I sang Mannam frequently, too.’

K: Na-nun [VP cacwu mannam-ul tul]-ess-e. Ne-nun? I-top frequently Mannam-acc listen-past-decl you-top ‘I listened to Mannam frequently. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I listened to Mannam frequently, too.’

209 (18) Kyengho-wa Hakpem-ika kakca etten kwail-ul cacwu sa-ss-nunci Kyengho-and Hakpem-nom each which fruit-acc frequently buy-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Kyengho and Hakpem were talking about which fruit they each bought frequently.’

K: Na-nun [VP cacwu sakwa-lul sa]-ss-e. Ne-nun? I-top frequently apple-acc buy-past-decl you-top ‘I bought apples frequently. How about you?’ H: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I bought apples frequently, too.’

H: Na-nun [VP mayil thayngko-lul chwu]-ess-e. Ne-nun? I-top every.day Tango-acc dance-past-decl you-top ‘I danced Tango everyday. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I danced Tango everyday, too.’

C: Na-nun [VP olay kay-lul khiwu]-ess-e. Ne-nun? I-top long dog-acc raise-past-decl you-top ‘I kept a dog for a long period of time. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I kept a dog for a long period of time, too.’

210 (21) Henswu-wa Canghwun-ika kakca etten mwulken-ul olay mo-ass-nunci Henswu-and Canghwun-nom each which item-acc long collect-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Henswu and Canghwun were talking about which item they each collected for a long period of time.’

H: Na-nun [VP olay wuphyo-lul mo]-ass-e. Ne-nun? I-top long stamp-acc collect-past-decl you-top ‘I collected stamps for a long period of time. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I collected stamps for a long period of time, too.’

C: Na-nun [VP olay pwule-lul paywu]-ess-e. Ne-nun? I-top long French-acc learn-past-decl you-top ‘I learned French for a long period of time. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I learned French for a long period of time, too.’

C: Na-nun [VP cacwu manulppang-ul kwu]-ess-e. Ne-nun? I-top frequently garlic.bread-acc bake-past-decl you-top ‘I baked garlic bread frequently. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I baked garlic bread frequently, too.’

211 (24) Tonghay-wa Siwan-ika kakca etten chayk-ul cacwu ilk-ess-nunci Tonghay-and Siwan-nom each which book-acc frequently read-past-comp malha-ko iss-ess-ta. talk-prog past-decl ‘Tonghay and Siwan were talking about which book they each read frequently.’

T: Na-nun [VP cacwu hayliphothe-lul ilk]-ess-e. Ne-nun? I-top frequently Harry.Potter-acc read-past-decl you-top ‘I read Harry Potter frequently. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I read Harry Potter frequently, too.’

E.3 Condition 3: Complex-Extraction (long distance scrambling)

(1) Minswu-wa Hanswu-ka kakca etten kwamok-ul Yuli-ka yelsimhi Minswu-and Hanswu-nom each which subject-acc Yuli-nom hard kongpwuha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. study-past-comp hear-past-comp talk-prog past-decl ‘Minswu and Hanswu were talking about which subject they each heard that Yuli studied hard.’

M: Swuhak-ul1 na-nun [VP[CP Yuli-ka yelsimhi t1 kongpwuha-yss-tako] math-acc I-top Yuli-nom hard study-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli studied math hard. How about you?’ H: Swuhak-ul na-to kulay-ss-e. math-acc I-also do.so-past-decl Intended: ‘I also heard that Yuli studied math hard.’

(2) Pongswu-wa Hyentong-ika kakca etten kkoch-ul Yuli-ka yelsimhi Pongswu-and Hyentong-nom each which ﬂower-acc Yuli-nom diligently sim-ess-tako tul-ess-nunci malha-ko iss-ess-ta. plant-past-comp hear-past-comp talk-prog past-decl ‘Pongswu and Hyentong were talking about which ﬂowers they each heard that Yuli planted diligently.’

P: Haypalaki-lul1 na-nun [VP[CP Yuli-ka yelsimhi t1 sim-ess-tako] sunﬂower-acc I-top Yuli-nom diligently plant-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

212 ‘I heard that Yuli planted sunflowers diligently. How about you?’ H: Haypalaki-lul na-to kulay-ss-e. sunflower-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli planted sunflowers diligently.’

(3) Sungho-wa Changcin-ika kakca etten cencaceyphwum-ul Yuli-ka olay Sungho-and Changcin-nom each which electronic.goods-acc Yuli-nom long sayongha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. use-past-comp hear-past-comp talk-prog past-decl ‘Sungho and Changcin were talking about which electronic goods they each heard that Yuli used for a long period of time.’

S: Nothupwuk-ul1 na-nun [VP[CP Yuli-ka olay t1 sayongha-yss-tako] laptop-acc I-top Yuli-nom long use-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli used a laptop for a long period of time. How about you?’ C: Nothupwuk-ul na-to kulay-ss-e. laptop-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli used a laptop for a long period of time.’

(4) Kyuho-wa Sangchel-ika kakca etten mwuswul-ul Yuli-ka olay Kyuho-and Sangchel-nom each which martial.arts-acc Yuli-nom long swulyenha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. practice-past-comp hear-past-comp talk-prog past-decl ‘Kyuho and Sangchel were talking about which martial arts they each heard that Yuli practiced for a long period of time.’

K: Thaykwuento-lul1 na-nun [VP[CP Yuli-ka olay t1 swulyenha-yss-tako] Taekwondo-acc I-top Yuli-nom long practice-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli practiced Taekwondo for a long period of time. How about you?’ S: Thaykwuento-lul na-to kulay-ss-e. Taekwondo-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli practiced Taekwondo for a long period of time.’

213 ‘Kyengphyo and Caychel were talking about which belongings they each heard that Yuli lost frequently.’

K: Cikap-ul1 na-nun [VP[CP Yuli-ka cacwu t1 ilepeli-ess-tako] wallet-acc I-top Yuli-nom frequently lose-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli lost a wallet freqeuntly. How about you?’ C: Cikap-ul na-to kulay-ss-e. wallet-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli lost a wallet frequently.’

(6) Congswu-wa Saypem-ika kakca etten cha-ul Yuli-ka cengmal Congswu-and Saypem-nom each which car-acc Yuli-nom really coaha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. like-past-comp hear-past-comp talk-prog past-decl ‘Congswu and Saypem were talking about which car they each heard that Yuli really liked.’

C: Nywupithul-ul1 na-nun [VP[CP Yuli-ka cengmal t1 coaha-yss-tako] New.Beetle-acc I-top Yuli-nom really like-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli really liked New Beetle. How about you?’ S: Nywupithul-ul na-to kulay-ss-e. New.Beetle-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli really liked New Beetle.’

(7) Swuho-wa Cayhak-ika kakca etten panchan-ul Yuli-ka cengmal Swuho-and Cayhak-nom each which side.dish-acc Yuli-nom really sileha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. dislike-past-comp hear-past-comp talk-prog past-decl ‘Swuho and Cayhak were talking about which side dish they each heard that Yuli really disliked.’

S: Kimchi-lul1 na-nun [VP[CP Yuli-ka cengmal t1 sileha-yss-tako] Kimchi-acc I-top Yuli-nom really dislike-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli really disliked Kimchi. How about you?’ C: Kimchi-lul na-to kulay-ss-e. Kimchi-acc I-also so.do-past-decl

214 Intended: ‘I also heard that Yuli really disliked Kimchi.’

(8) Congwu-wa Taykwan-ika kakca etten homsyophing sangphwum-ul Congwu-and Taykwan-nom each which home.shopping goods-acc Yuli-ka cacwu cwumwunha-yss-tako tul-ess-nunci malha-ko Yuli-nom frequently order-past-comp hear-past-comp talk-prog iss-ess-ta. past-decl ‘Congwu and Taykwan were talking about which home shopping goods they each heard that Yuli ordered frequently.’

C: Kimchi-lul1 na-nun [VP[CP Yuli-ka cacwu t1 cwumwunha-yss-tako] Kimchi-acc I-top Yuli-nom frequently order-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli ordered Kimchi frequently. How about you?’ T: Kimchi-lul na-to kulay-ss-e. Kimchi-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli ordered Kimchi frequently.’

(9) Bemswu-wa Twusik-ika kakca etten thipiphulo-lul Yuli-ka Bemswu-and Twusik-nom each which television.program-acc Yuli-nom cacwu sichengha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. frequently watch-past-comp hear-past-comp talk-prog past-decl ‘Bemswu and Twusik were talking about which TV show they each heard that Yuli watched frequently.’

B: Mwuhantocen-ul1 na-nun [VP[CP Yuli-ka cacwu t1 Infinite.Challenge-acc I-top Yuli-nom frequently sichengha-yss-tako] tul]-ess-e. Ne-nun? watch-past-comp hear-past-decl you-top ‘I heard that Yuli watched Infinite Challenge frequently. How about you?’ T: Mwuhantocen-ul na-to kulay-ss-e. Infinite.Challenge-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli watched Infinite Challenge frequently.’

(10) Chelswu-wa Heechel-ika kakca etten maykcwu-lul Yuli-ka cacwu Chelswu-and Heechel-nom each which beer-acc Yuli-nom frequently masi-ess-tako tul-ess-nunci malha-ko iss-ess-ta. drink-past-comp hear-past-comp talk-prog past-decl ‘Chelswu and Heechel were talking about which beer they each heard that Yuli drank frequently.’

215 C: Chasu-lul1 na-nun [VP[CP Yuli-ka cacwu t1 masi-ess-tako] Cass-acc I-top Yuli-nom frequently drink-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli drank Cass frequently. How about you?’ H: Chasu-lul na-to kulay-ss-e. Cass-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli drank Cass frequently.’

(11) Minswu-wa Yengmin-ika kakca etten ilponumsik-ul Yuli-ka Minswu-and Yengmin-nom each which Japanese.food-acc Yuli-nom cacwu mek-ess-tako tul-ess-nunci malha-ko iss-ess-ta. frequently eat-past-comp hear-past-comp talk-prog past-decl ‘Minswu and Yengmin were talking about which Japanese food they each heard that Yuli ate frequently.’

M: Udong-ul1 na-nun [VP[CP Yuli-ka cacwu t1 mek-ess-tako] Udon-acc I-top Yuli-nom frequently eat-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli ate Udon frequently. How about you?’ Y: Udong-ul na-to kulay-ss-e. Udon-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli ate Udon frequently.’

(12) Sengmo-wa Cinchel-ika kakca etten sinmwun-ul Yuli-ka olay Sengmo-and Cinchel-nom each which newspaper-acc Yuli-nom long kwutokha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. subscribe-past-comp hear-past-comp talk-prog past-decl ‘Sengmo and Cinchel were talking about which newspaper they each heard that Yuli subscribed to for a long period of time.’

S: Tongailpo-lul1 na-nun [VP[CP Yuli-ka olay t1 kwutokha-yss-tako] Dong-A.Ilbo-acc I-top Yuli-nom long subscribe-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli subscribed to Dong-A Ilbo for a long period of time. How about you?’ C: Tongailpo-lul na-to kulay-ss-e. DongA.Ilbo-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli subscribed to Dong-A Ilbo for a long period of time.’

216 (13) Hyenho-wa Tayhwun-ika kakca etten yenghwa-lul Yuli-ka cacwu Hyenho-and Tayhwun-nom each which movie-acc Yuli-nom frequently poa-ss-tako tul-ess-nunci malha-ko iss-ess-ta. watch-past-comp hear-past-comp talk-prog past-decl ‘Hyenho and Tayhwun were talking about which movie they each heard that Yuli watched frequently.’

H: Phulocun-ul1 na-nun [VP[CP Yuli-ka cacwu t1 poa-ss-tako] Frozen-acc I-top Yuli-nom frequently watch-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli watched Frozen frequently. How about you?’ T: Phulocun-ul na-to kulay-ss-e. Frozen-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli watched Frozen frequently.’

(14) Thayswu-wa Myenghwun-ika kakca etten akki-lul Yuli-ka Thayswu-and Myenghwun-nom each which music.instrument-acc Yuli-nom olay yencwuha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. long play-past-comp hear-past-comp talk-prog past-decl ‘Thayswu and Myenghwun were talking about which music instrument they each heard that Yuli played for a long period of time.’

T: Phiano-lul1 na-nun [VP[CP Yuli-ka olay t1 yencwuha-yss-tako] piano-acc I-top Yuli-nom long play-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli played the piano for a long period of time. How about you?’ M: Phiano-lul na-to kulay-ss-e. piano-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli played the piano for a long period of time.’

(15) Hungswu-wa Minhyen-ika kakca etten hankwukumsik-ul Yuli-ka Hungswu-and Minhyen-nom each which Korean.food-acc Yuli-nom cacwu yoliha-yss-tako tul-ess-nunci malha-ko iss-ess-ta. frequently cook-past-comp hear-past-comp talk-prog past-decl ‘Hungswu and Minhyen were talking about which Korean food they each heard that Yuli cooked frequently.’

H: Kimchi-lul1 na-nun [VP[CP Yuli-ka cacwu t1 yoliha-yss-tako] Kimchi-acc I-top Yuli-nom frequently cook-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

217 ‘I heard that Yuli cooked Kimchi frequently. How about you?’ M: Kimchi-lul na-to kulay-ss-e. Kimchi-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli cooked Kimchi frequently.’

(16) Unphyo-wa Congsin-ika kakca etten nolay-lul Yuli-ka cacwu Unphyo-and Congsin-nom each which song-acc Yuli-nom frequently pwulu-ess-tako tul-ess-nunci malha-ko iss-ess-ta. sing-past-comp hear-past-comp talk-prog past-decl ‘Unphyo and Congsin were talking about which song they each heard that Yuli sang frequently.’

U: Mannam-ul1 na-nun [VP[CP Yuli-ka cacwu t1 pwulu-ess-tako] Mannam-acc I-top Yuli-nom frequently sing-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli sang Mannam frequently. How about you?’ C: Mannam-ul na-to kulay-ss-e. Mannam-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli sang Mannam frequently.’

(17) Kithay-wa Cenghwun-ika kakca etten nolay-lul Yuli-ka cacwu Kithay-and Cenghwun-nom each which song-acc Yuli-nom frequently tul-ess-tako tul-ess-nunci malha-ko iss-ess-ta. listen-past-comp hear-past-comp talk-prog past-decl ‘Kithay and Cenghwun were talking about which song they each heard that Yuli listened to frequently.’

K: Mannam-ul1 na-nun [VP[CP Yuli-ka cacwu t1 tul-ess-tako] Mannam-acc I-top Yuli-nom frequently listen-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli listened to Mannam frequently. How about you?’ C: Mannam-ul na-to kulay-ss-e. Mannam-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli listened to Mannam frequently.’

(18) Kyengho-wa Hakpem-ika kakca etten kwail-ul Yuli-ka cacwu Kyengho-and Hakpem-nom each which fruit-acc Yuli-nom frequently sa-ss-tako tul-ess-nunci malha-ko iss-ess-ta. buy-past-comp hear-past-comp talk-prog past-decl ‘Kyengho and Hakpem were talking about which fruit they each heard that Yuli bought frequently.’

218 K: Sakwa-lul1 na-nun [VP[CP Yuli-ka cacwu t1 sa-ss-tako] apple-acc I-top Yuli-nom frequently buy-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli bought apples frequently. How about you?’ H: Sakwa-lul na-to kulay-ss-e. apple-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli bought apples frequently.’

(19) Hyenwu-wa Tonghwun-ika kakca etten chwum-ul Yuli-ka mayil Hyenwu-and Tonghwun-nom each which dance-acc Yuli-nom every.day chwu-ess-tako tul-ess-nunci malha-ko iss-ess-ta. dance-past-comp hear-past-comp talk-prog past-decl ‘Hyenwu and Tonghwun were talking about which dance they each heard that Yuli danced everyday.’

H: Thayngko-lul1 na-nun [VP[CP Yuli-ka mayil t1 chwu-ess-tako] Tango-acc I-top Yuli-nom every.day dance-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli danced Tango everyday. How about you?’ T: Thayngko-lul na-to kulay-ss-e. Tango-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli danced Tango everyday.’

(20) Ciseng-iwa Mincong-ika kakca etten aywantongmwul-ul Yuli-ka olay Ciseng-and Mincong-nom each which pet-acc Yuli-nom long khiwu-ess-tako tul-ess-nunci malha-ko iss-ess-ta. raise-past-comp hear-past-comp talk-prog past-decl ‘Ciseng and Mincong were talking about which pet they each heard that Yuli kept for a long period of time.’

C: Kay-lul1 na-nun [VP[CP Yuli-ka olay t1 khiwu-ess-tako] dog-acc I-top Yuli-nom long raise-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli kept a dog for a long period of time. How about you?’ M: Kay-lul na-to kulay-ss-e. dog-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli kept a dog for a long period of time.’

219 (21) Henswu-wa Canghwun-ika kakca etten mwulken-ul Yuli-ka olay Henswu-and Canghwun-nom each which item-acc Yuli-nom long mo-ass-tako tul-ess-nunci malha-ko iss-ess-ta. collect-past-comp hear-past-comp talk-prog past-decl ‘Henswu and Canghwun were talking about which item they each heard that Yuli collected for a long period of time.’

H: Wuphyo-lul1 na-nun [VP[CP Yuli-ka olay t1 mo-ass-tako] stamp-acc I-top Yuli-nom long collect-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli collected stamps for a long period of time. How about you?’ C: Wuphyo-lul na-to kulay-ss-e. stamp-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli collected stamps for a long period of time.’

(22) Changho-wa Minhyen-ika kakca etten wuekuke-lul Yuli-ka olay Changho-and Minhyen-nom each which foreign.language-acc Yuli-nom long paywu-ess-tako tul-ess-nunci malha-ko iss-ess-ta. learn-past-comp hear-past-comp talk-prog past-decl ‘Changho and Minhyen were talking about which foreign language they each heard that Yuli learned for a long period of time.’

C: Pwule-lul1 na-nun [VP[CP Yuli-ka olay t1 paywu-ess-tako] French-acc I-top Yuli-nom long learn-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli learned French for a long period of time. How about you?’ M: Pwule-lul na-to kulay-ss-e. French-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli learned French for a long period of time.’

(23) Cengho-wa Senghyen-ika kakca etten ppang-ul Yuli-ka cacwu Cengho-and Senghyen-nom each which bread-acc Yuli-nom frequently kwu-ess-tako tul-ess-nunci malha-ko iss-ess-ta. bake-past-comp hear-past-comp talk-prog past-decl ‘Cengho and Senghyen were talking about which bread they each heard that Yuli baked frequently.’

C: Manulppang-ul1 na-nun [VP[CP Yuli-ka cacwu t1 kwu-ess-tako] garlic.bread-acc I-top Yuli-nom frequently bake-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

220 ‘I heard that Yuli baked garlic bread frequently. How about you?’ S: Manulppang-ul na-to kulay-ss-e. garlic.bread-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli baked garlic bread frequently, too.’

(24) Tonghay-wa Siwan-ika kakca etten chayk-ul Yuli-ka cacwu Tonghay-and Siwan-nom each which book-acc Yuli-nom frequently ilk-ess-tako tul-ess-nunci malha-ko iss-ess-ta. read-past-comp hear-past-comp talk-prog past-decl ‘Tonghay and Siwan were talking about which book they each heard that Yuli read frequently.’

T: Hayliphothe-lul1 na-nun [VP[CP Yuli-ka cacwu t1 ilk-ess-tako] Harry.Potter-acc I-top Yuli-nom frequently read-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli read Harry Potter frequently. How about you?’ S: Hayliphothe-lul na-to kulay-ss-e. Harry.Potter-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli read Harry Potter frequently.’

E.4 Condition 4: Complex-NoExtraction

(1) Minswu-wa Hanswu-ka kakca etten kwamok-ul Yuli-ka yelsimhi Minswu-and Hanswu-nom each which subject-acc Yuli-nom hard kongpwuha-yess-tako sayngkakha-yess-nunci malha-ko iss-ess-ta. study-past-comp think-past-comp talk-prog past-decl ‘Minswu and Hanswu were talking about which subject they each thought that Yuli studied hard.’

M: Na-nun [VP[CP Yuli-ka yelsimhi swuhak-ul kongpwuha-yss-tako] I-top Yuli-nom hard math-acc study-past-comp sayngkakha]-yss-e. Ne-nun? think-past-decl you-top ‘I thought that Yuli studied math hard. How about you?’ H: Na-to kulay-ss-e. I-also do.so-past-decl Intended: ‘I also thought that Yuli studied math hard.’

221 ‘Pongswu and Hyentong were talking about which ﬂowers they each heard that Yuli planted diligently.’

P: Na-nun [VP[CP Yuli-ka yelsimhi haypalaki-lul sim-ess-tako] I-top Yuli-nom diligently sunflower-acc plant-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli planted sunflowers diligently. How about you?’ H: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli planted sunflowers diligently.’

S: Na-nun [VP[CP Yuli-ka olay nothupwuk-ul sayongha-yss-tako] I-top Yuli-nom long laptop-acc use-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli used a laptop for a long period of time. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli used a laptop for a long period of time.’

K: Na-nun [VP[CP Yuli-ka olay thaykwuento-lul swulyenha-yss-tako] I-top Yuli-nom long Taekwondo-acc practice-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli practiced Taekwondo for a long period of time. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl

222 Intended: ‘I also heard that Yuli practiced Taekwondo for a long period of time.’

(5) Kyengphyo-wa Caychel-ika kakca etten sociphwum-ul Yuli-ka cacwu Kyengphyo-and Caychel-nom each which belongings-acc Yuli-nom frequently ilepeli-ess-tako tul-ess-nunci malha-ko iss-ess-ta. lose-past-comp hear-past-comp talk-prog past-decl ‘Kyengphyo and Caychel were talking about which belongings they each heard that Yuli lost frequently.’

K: Na-nun [VP[CP Yuli-ka cacwu cikap-ul ilepeli-ess-tako] I-top Yuli-nom frequently wallet-acc lose-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli lost a wallet freqeuntly. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli lost a wallet frequently.’

C: Na-nun [VP[CP Yuli-ka cengmal nywupithul-ul coaha-yss-tako] I-top Yuli-nom really New.Beetle-acc like-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli really liked New Beetle. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli really liked New Beetle.’

223 S: Na-nun [VP[CP Yuli-ka cengmal kimchi-lul sileha-yss-tako] I-top Yuli-nom really Kimchi-acc dislike-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli really disliked Kimchi. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli really disliked Kimchi.’

C: Na-nun [VP[CP Yuli-ka cacwu kimchi-lul cwumwunha-yss-tako] I-top Yuli-nom frequently Kimchi-acc order-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli ordered Kimchi frequently. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli ordered Kimchi frequently.’

B: Na-nun [VP[CP Yuli-ka cacwu mwuhantocen-ul I-top Yuli-nom frequently Infinite.Challenge-acc sichengha-yss-tako] tul]-ess-e. Ne-nun? watch-past-comp hear-past-decl you-top ‘I heard that Yuli watched Infinite Challenge frequently. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli watched Infinite Challenge frequently.’

224 (10) Chelswu-wa Heechel-ika kakca etten maykcwu-lul Yuli-ka cacwu Chelswu-and Heechel-nom each which beer-acc Yuli-nom frequently masi-ess-tako tul-ess-nunci malha-ko iss-ess-ta. drink-past-comp hear-past-comp talk-prog past-decl ‘Chelswu and Heechel were talking about which beer they each heard that Yuli drank frequently.’

C: Na-nun [VP[CP Yuli-ka cacwu chasu-lul masi-ess-tako] I-top Yuli-nom frequently Cass-acc drink-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli drank Cass frequently. How about you?’ H: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli drank Cass frequently.’

M: Na-nun [VP[CP Yuli-ka cacwu udong-ul mek-ess-tako] I-top Yuli-nom frequently Udon-acc eat-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli ate Udon frequently. How about you?’ Y: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli ate Udon frequently.’

S: Na-nun [VP[CP Yuli-ka olay tongailpo-lul kwutokha-yss-tako] I-top Yuli-nom long Dong-A.Ilbo-acc subscribe-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

225 ‘I heard that Yuli subscribed to Dong-A Ilbo for a long period of time. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli subscribed to Dong-A Ilbo for a long period of time.’

(13) Hyenho-wa Tayhwun-ika kakca etten yenghwa-lul Yuli-ka cacwu Hyenho-and Tayhwun-nom each which movie-acc Yuli-nom frequently poa-ss-tako tul-ess-nunci malha-ko iss-ess-ta. watch-past-comp hear-past-comp talk-prog past-decl ‘Hyenho and Tayhwun were talking about which movie they each heard that Yuli watched frequently.’

H: Na-nun [VP[CP Yuli-ka cacwu phulocun-ul poa-ss-tako] I-top Yuli-nom frequently Frozen-acc watch-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli watched Frozen frequently. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli watched Frozen frequently.’

T: Na-nun [VP[CP Yuli-ka olay phiano-lul yencwuha-yss-tako] I-top Yuli-nom long piano-acc play-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli played the piano for a long period of time. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli played the piano for a long period of time.’

226 ‘Hungswu and Minhyen were talking about which Korean food they each heard that Yuli cooked frequently.’

H: Na-nun [VP[CP Yuli-ka cacwu kimchi-lul yoliha-yss-tako] I-top Yuli-nom frequently Kimchi-acc cook-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli cooked Kimchi frequently. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli cooked Kimchi frequently.’

U: Na-nun [VP[CP Yuli-ka cacwu mannam-ul pwulu-ess-tako] I-top Yuli-nom frequently Mannam-acc sing-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli sang Mannam frequently. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli sang Mannam frequently.’

K: Na-nun [VP[CP Yuli-ka cacwu mannam-ul tul-ess-tako] I-top Yuli-nom frequently Mannam-acc listen-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli listened to Mannam frequently. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl

227 Intended: ‘I also heard that Yuli listened to Mannam frequently.’

K: Na-nun [VP[CP Yuli-ka cacwu sakwa-lul sa-ss-tako] I-top Yuli-nom frequently apple-acc buy-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli bought apples frequently. How about you?’ H: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli bought apples frequently.’

H: Na-nun [VP[CP Yuli-ka mayil thayngko-lul chwu-ess-tako] I-top Yuli-nom every.day Tango-acc dance-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli danced Tango everyday. How about you?’ T: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli danced Tango everyday.’

C: Na-nun [VP[CP Yuli-ka olay kay-lul khiwu-ess-tako] I-top Yuli-nom long dog-acc raise-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top

228 ‘I heard that Yuli kept a dog for a long period of time. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli kept a dog for a long period of time.’

(21) Henswu-wa Canghwun-ika kakca etten mwulken-ul Yuli-ka olay Henswu-and Canghwun-nom each which item-acc Yuli-nom long mo-ass-tako tul-ess-nunci malha-ko iss-ess-ta. collect-past-comp hear-past-comp talk-prog past-decl ‘Henswu and Canghwun were talking about which item they each heard that Yuli collected for a long period of time.’

H: Na-nun [VP[CP Yuli-ka olay wuphyo-lul mo-ass-tako] I-top Yuli-nom long stamp-acc collect-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli collected stamps for a long period of time. How about you?’ C: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli collected stamps for a long period of time.’

C: Na-nun [VP[CP Yuli-ka olay pwule-lul paywu-ess-tako] I-top Yuli-nom long French-acc learn-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli learned French for a long period of time. How about you?’ M: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli learned French for a long period of time.’

229 C: Na-nun [VP[CP Yuli-ka cacwu manulppang-ul kwu-ess-tako] I-top Yuli-nom frequently garlic.bread-acc bake-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli baked garlic bread frequently. How about you?’ S: Manulppang-ul na-to kulay-ss-e. garlic.bread-acc I-also so.do-past-decl Intended: ‘I also heard that Yuli baked garlic bread frequently, too.’

T: Na-nun [VP[CP Yuli-ka cacwu hayliphothe-lul ilk-ess-tako] I-top Yuli-nom frequently Harry.Potter-acc read-past-comp tul]-ess-e. Ne-nun? hear-past-decl you-top ‘I heard that Yuli read Harry Potter frequently. How about you?’ S: Na-to kulay-ss-e. I-also so.do-past-decl Intended: ‘I also heard that Yuli read Harry Potter frequently.’

230 Appendix F

Test sentences from Experiment 5

F.1 Condition 1: NullObject-Bound (sloppy identity reading)

(1) Minswu, Kiswu, Cinswu-ka say kiswuksa-lo isaha-ko iss-ess-ta. Minswu Kiswu Cinswu-nom new dormitory-to move-prog past-decl Minswu-ka Minswu-uy cim-ul nalu-ess-ta. Kiswu-to Kiswu-uy Minswu-nom Minswu-gen stuff-acc move-past-decl Kiswu-also Kiswu-gen cim-ul nalu-ess-ta. stuff-acc move-past-decl ‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuff. Kiswu also moved Kiswu’s stuff.’ Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] Minswu-nom he-gen stuff-acc move-past-conj Kiswu-also nalu-ess-ta. move-past-decl ‘Minswu moved his stuff, and Kiswu moved, too.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse yakwu-lul ha-ko Yonghi Kenhi Myengho-nom play.ground-in baseball-acc do-prog iss-ess-ta. Yonghi-ka Yonghi-uy kong-ul tenci-ess-ta. Kenhi-to past-decl Yonghi-nom Yonghi-gen ball-acc throw-past-decl Kenhi-also Kenhi-uy kong-ul tenci-ess-ta. Kenhi-gen ball-acc throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground. Yonghi threw Yonghi’s ball. Kenhi also threw Kenhi’s ball.’ Yonghi-ka ku-uy kong-ul tenci-ess-ko, Kenhi-to [e] Yonghi-nom he-gen ball-acc throw-past-conj Kenhi-also tenci-ess-ta. throw-past-decl

231 ‘Yonghi threw his ball, and Kenhi threw, too.’

(3) Thayswu, Kiswu, Cengswu-ka pa-eyse wuain-ul masi-ko iss-ess-ta. Thayswu Kiswu Cengswu-nom bar-at wuain-acc drink-prog past-decl Thayswu-ka Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-to Thayswu-nom Thayswu-gen wine.glass-acc break-past-decl Kiswu-also Kiswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-gen wine.glass-acc break-past-decl ‘Thayswu, Kiswu, and Cengswu were drinking wine at a bar. Thayswu broke Thayswu’s wine glass. Kiswu also broke Kiswu’s wine glass.’ Thayswu-ka ku-uy wuaincan-ul kkaythuli-ess-ko, Kiswu-to [e] Thayswu-nom he-gen wine.glass-acc break-past-conj Kiswu-also kkaythuli-ess-ta. break-past-decl ‘Thayswu broke his wine glass, and Kiswu broke, too.’

(4) Myengswu, Hyenswu, Cwunswu-ka wuntong hwu swui-ko iss-ess-ta. Myengswu Hyenswu Cwunswu-nom exercise after rest-prog past-decl Myengswu-ka Myengswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-to Myengswu-nom Myengswu-gen beverage-acc drink-past-decl Hyenswu-also Hyenswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-gen beverage-acc drink-past-decl ‘Myengswu, Hyenswu, and Cwunswu were taking a rest after exercise. Myengswu drank Myengswu’s beverage. Hyenswu also drank Hyenswu’s beverage.’ Myengswu-ka ku-uy umlyoswu-lul masi-ess-ko, Hyenswu-to [e] Myengswu-nom he-gen beverage-acc drink-past-conj Hyenswu-also masi-ess-ta. drink-past-decl ‘Myengswu drank his beverage, and Hyenswu drank, too.’

(5) Pyengswu, Mwunswu, Cwunswu-ka cip-ul chengsoha-ko iss-ess-ta. Pyengswu Mwunswu Cwunswu-nom house-acc clean-prog past-decl Pyengswu-ka Pyengswu-uy pang-ul ssul-ess-ta. Mwunswu-to Pyengswu-nom Pyengswu-gen room-acc sweep-past-decl Mwunswu-also Mwunswu-uy pang-ul ssul-ess-ta. Mwunswu-gen room-acc sweep-past-decl ‘Pyengswu, Mwunswu, and Cwunswu were cleaning their house. Pyengswu swept Pyengswu’s room. Mwunswu also swept Mwunswu’s room.’ Pyengswu-ka ku-uy pang-ul ssul-ess-ko, Mwunswu-to [e] Pyengswu-nom he-gen room-acc sweep-past-conj Mwunswu-also ssul-ess-ta. sweep-past-decl

232 ‘Pyengswu swept his room, and Mwunswu swept, too.’

(6) Chelswu, Yengswu, Cengswu-ka tosekwuan-eyse kongpwuha-ko iss-ess-ta. Chelswu Yengswu Cengswu-nom library-in study-prog past-decl Chelswu-ka Chelswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-to Chelswu-nom Chelswu-gen pen-acc drop-past-decl Yengswu-also Yengswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-gen pen-acc drop-past-decl ‘Chelswu, Yengswu, and Cengswu were studying in a library. Chelswu dropped Chelswu’s pen. Yengswu also dropped Yengswu’s pen.’ Chelswu-ka ku-uy pheyn-ul ttelethuli-ess-ko, Yengswu-to [e] Chelswu-nom he-gen pen-acc drop-past-conj Chelswu-also ttelethuli-ess-ta. drop-past-decl ‘Chelswu dropped his pen, and Yengswu dropped, too.’

(7) Chelho, Phengho, Swuho-ka kakca phyenci-lul ssu-ko iss-ess-ta. Chelho Phengho Swuho-nom each letter-acc write-prog past-decl Chelho-ka Chelho-uy phyenci-lul ilk-ess-ta. Phengho-to Chelho-nom Chelho-gen letter-acc read-past-decl Phengho-also Phengho-uy phyenci-lul ilk-ess-ta. Phengho-gen letter-acc read-past-decl ‘Chelho, Phengho, and Swuho were each writing a letter. Chelho read Chelho’s letter. Phengho also read Phengho’s letter.’ Chelho-ka ku-uy phyenci-lul ilk-ess-ko, Phengho-to [e] Chelho-nom he-gen letter-acc read-past-conj Phengho-also ilk-ess-ta. read-past-decl ‘Chelho read his letter, and Phengho read, too.’

(8) Kwangho, Changho, Yengho-ka ilsiktang-eyse cemsim-ul mek-ko Kwangho Changho Yengho-nom Japanese.restaurant-at lunch-acc eat-prog iss-ess-ta. Kwangho-ka Kwangho-uy utong-ul mek-ess-ta. past-decl Kwangho-nom Kwangho-gen Udon-acc eat-past-decl Changho-to Changho-uy utong-ul mek-ess-ta. Changho-also Changho-gen Udon-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having lunch at a Japanese restaurant. Kwangho ate Kwangho’s Udon. Changho also ate Changho’s Udon.’ Kwangho-ka ku-uy utong-ul mek-ess-ko, Changho-to [e] Kwangho-nom he-gen Udon-acc eat-past-conj Changho-also mek-ess-ta. eat-past-decl

233 ‘Kwangho ate his Udon, and Changho ate, too.’

(9) Pyengmin, Cengmin, Tongmin-ika wuntongcang-eyse chwukkwu-lul ha-ko Pyengmin Cengmin Tongmin-nom play.ground-in soccer-acc do-prog iss-ess-ta. Pyengmin-ika Pyengmin-iuy kong-ul cha-ss-ta. past-decl Pyengmin-nom Pyengmin-gen ball-acc kick-past-decl Cengmin-ito Cengmin-iuy kong-ul cha-ss-ta. Cengmin-also Cengmin-gen ball-acc kick-past-decl ‘Pyengmin, Cengmin, and Tongmin were playing soccer in a playground. Pyeng- min kicked Pyengmin’s ball. Cengmin also kicked Cengmin’s ball.’ Pyengmin-ika ku-uy kong-ul cha-ss-ko, Cengmin-ito [e] Pyengmin-nom he-gen ball-acc kick-past-conj Cengmin-also cha-ss-ta. kick-past-decl ‘Pyengmin kicked his ball, and Cengmin kicked, too.’

(10) Cinswu, Sungswu, Pyengswu-ka seythaksil-eyse seythak-ul ha-ko Cinswu Sungswu Pyengswu-nom laundry.room-at laundry-acc do-prog iss-ess-ta. Cinswu-ka Cinswu-uy seycey-lul sayongha-yss-ta. past-decl Cinswu-nom Cinswu-gen detergent-acc use-past-decl Sungswu-to Sungswu-uy seycey-lul sayongha-yss-ta. Sungswu-also Sungswu-gen detergent-acc use-past-decl ‘Cinswu, Sungswu, and Pyengswu were doing laundry at a laundry room. Cinswu used Cinswu’s detergent. Sungswu also used Sungswu’s detergent.’ Cinswu-ka ku-uy seycey-lul sayongha-yss-ko, Sungswu-to [e] Cinswu-nom he-gen detergent-acc use-past-conj Sungswu-also sayongha-yss-ta. use-past-decl ‘Cinswu used his detergent, and Sungswu used, too.’

(11) Kimkenmo, Kimcanghwun, Sinsunghwun-i nolaypang-eyse nol-ko Kimkenmo Kimcanghwun Sinsunghwun-nom singing.room-at hang.out-prog iss-ess-ta. Kimkenmo-ka Kimkenmo-uy hithukok-ul pwulu-ess-ta. past-decl Kimkenmo-nom Kimkenmo-gen hit.song-acc sing-past-decl Kimcanghwun-to Kimcanghwun-uy hithukok-ul pwulu-ess-ta. Kimcanghwun-also Kimcanghwun-gen hit.song-acc sing-past-decl ‘Kimkenmo, Kimcanghwun, and Sinsunghwun were hanging out in a singing room. Kimkenmo sang Kimkenmo’s hit song. Kimcanghwun also sang Kim- canghwun’s hit song.’ Kimkenmo-ka ku-uy hithukok-ul pwulu-ess-ko, Kimcanghwun-to [e] Kimkenmo-nom he-gen hit.song-acc sing-past-conj Kimcanghwun-also pwulu-ess-ta. sing-past-decl

234 ‘Kimkenmo sang his hit song, and Kimcanghwun sang, too.’

(12) Pongswu, Thaykang, Thayho-ka pwuek-eyse selkeci-lul ha-ko Pongswu Thaykang Thayho-nom kitchen-at dish.washing-acc do-prog iss-ess-ta. Pongswu-ka Pongswu-uy khep-ul sis-ess-ta. past-decl Pongswu-nom Pongswu-gen cup-acc wash-past-decl Thaykang-ito Thaykang-iuy khep-ul sis-ess-ta. Thaykang-also Thaykang-gen cup-acc wash-past-decl ‘Pongswu, Thaykang, and Thayho were doing the dishes at a kitchen. Pongswu washed Pongswu’s cup. Thaykang also washed Thaykang’s cup.’ Pongswu-ka ku-uy khep-ul sis-ess-ko, Thaykang-ito [e] Pongswu-nom he-gen cup-acc wash-past-conj Thaykang-also sis-ess-ta. wash-past-decl ‘Pongswu washed his cup, and Thaykang washed, too.’

(13) Chiswu, Pyengswu, Hanswu-ka kkaphey-eyse tayhwua-lul nanwu-ko Chiswu Pyengswu Hanswu-nom café-at conversation-acc share-prog iss-ess-ta. Chiswu-ka Chiswu-uy cen yecachinkwu-lul past-decl Chiswu-nom Chiswu-gen previous girlfriend-acc helttut-ess-ta. Pyengswu-to Pyengswu-uy cen yecachinkwu-lul speak.ill.of-past-decl Pyengswu-also Pyengswu-gen previous girlfriend-acc helttut-ess-ta. speak.ill.of-past-decl ‘Chiswu, Pyengswu, and Hanswu were having a conversation at a café. Chiswu spoke ill of Chiswu’s ex-girlfriend. Pyengswu also spoke ill of Pyengswu’s ex- girlfriend.’ Chiswu-ka ku-uy cen yecachinkwu-lul helttut-ess-ko, Chiswu-nom he-gen previous girlfriend-acc speak.ill.of-past-conj Pyengswu-to [e] helttut-ess-ta. Pyengswu-also speak.ill.of-past-decl ‘Chiswu spoke ill of his ex-girlfriend, and Pyengswu spoke ill of, too.’

235 Congswu-ka ku-uy kay-lul kancilephi-ess-ko, Hwuanswu-to [e] Congswu-nom he-gen dog-acc tickle-past-conj Hwuanswu-also kancilephi-ess-ta. tickle-past-decl ‘Congswu tickled his dog, and Hwuanswu tickled, too.’

(15) Tongwu, Cinwu, Thaywu-ka Sinswu-uy cip-eyse nol-ko Tongwu Cinwu Thaywu-nom Sinswu-gen house-at hang.out-prog iss-ess-ta. Tongwu-ka Tongwu-uy yetongsayng-ul ttayli-ess-ta. past-decl Tongwu-nom Tongwu-gen younger.sister-acc hit-past-decl Cinwu-to Cinwu-uy yetongsayng-ul ttayli-ess-ta. Cinwu-also Cinwu-gen younger.sister-acc hit-past-decl ‘Tongwu, Cinwu, and Thaywu were hanging out at Sinswu’s house. Tongwu hit Tongwu’s younger sister. Cinwu also hit Cinwu’s younger sister.’ Tongwu-ka ku-uy yetongsayng-ul ttayli-ess-ko, Cinwu-to [e] Tongwu-nom he-gen younger.sister-acc hit-past-conj Cinwu-also ttayli-ess-ta. hit-past-decl ‘Tongwu hit his younger sister, and Cinwu hit, too.’

(16) Wonmwu, Kiswu, Sengswu-ka chatcip-eyse tayhwua-lul nanwu-ko Wonmwu Kiswu Sengswu-nom tea.house-at conversation-acc share-prog iss-ess-ta. Wonmwu-ka Wonmwu-uy nwuna-lul past-decl Wonmwu-nom Wonmwu-gen older.sister-acc chingchanha-yss-ta. Kiswu-to Kiswu-uy nwuna-lul compliment-past-decl Kiswu-also Kiswu-gen older.sister-acc chingchanha-yss-ta. compliment-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a tea house. Won- mwu complimented Wonmwu’s older sister. Kiswu also complimented Kiswu’s older sister.’ Wonmwu-ka ku-uy nwuna-lul chingchanha-yss-ko, Kiswu-to [e] Wonmwu-nom he-gen older.sister-acc compliment-past-conj Kiswu-also chingchanha-yss-ta. compliment-past-decl ‘Wonmwu complimented his older sister, and Kiswu complimented, too.’

236 F.2 Condition 2: NullObject-Free (strict identity reading)

(1) Minswu, Kiswu, Cinswu-ka say kiswuksa-lo isaha-ko iss-ess-ta. Minswu Kiswu Cinswu-nom new dormitory-to move-prog past-decl Minswu-ka Minswu-uy cim-ul nalu-ess-ta. Kiswu-to Minswu-uy Minswu-nom Minswu-gen stuff-acc move-past-decl Kiswu-also Minswu-gen cim-ul nalu-ess-ta. stuff-acc move-past-decl ‘Minswu, Kiswu, and Cinswu were moving to a new dormitory. Minswu moved Minswu’s stuff. Kiswu also moved Minswu’s stuff.’ Minswu-ka ku-uy cim-ul nalu-ess-ko, Kiswu-to [e] Minswu-nom he-gen stuff-acc move-past-conj Kiswu-also nalu-ess-ta. move-past-decl ‘Minswu moved his stuff, and Kiswu moved, too.’

(2) Yonghi, Kenhi, Myengho-ka wuntongcang-eyse yakwu-lul ha-ko Yonghi Kenhi Myengho-nom play.ground-in baseball-acc do-prog iss-ess-ta. Yonghi-ka Yonghi-uy kong-ul tenci-ess-ta. Kenhi-to past-decl Yonghi-nom Yonghi-gen ball-acc throw-past-decl Kenhi-also Yonghi-uy kong-ul tenci-ess-ta. Yonghi-gen ball-acc throw-past-decl ‘Yonghi, Kenhi, and Myengho were playing basketball in a playground. Yonghi threw Yonghi’s ball. Kenhi also threw Yonghi’s ball.’ Yonghi-ka ku-uy kong-ul tenci-ess-ko, Kenhi-to [e] Yonghi-nom he-gen ball-acc throw-past-conj Kenhi-also tenci-ess-ta. throw-past-decl ‘Yonghi threw his ball, and Kenhi threw, too.’

(3) Thayswu, Kiswu, Cengswu-ka pa-eyse wuain-ul masi-ko iss-ess-ta. Thayswu Kiswu Cengswu-nom bar-at wuain-acc drink-prog past-decl Thayswu-ka Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Kiswu-to Thayswu-nom Thayswu-gen wine.glass-acc break-past-decl Kiswu-also Thayswu-uy wuaincan-ul kkaythuli-ess-ta. Thayswu-gen wine.glass-acc break-past-decl ‘Thayswu, Kiswu, and Cengswu were drinking wine at a bar. Thayswu broke Thayswu’s wine glass. Kiswu also broke Thayswu’s wine glass.’ Thayswu-ka ku-uy wuaincan-ul kkaythuli-ess-ko, Kiswu-to [e] Thayswu-nom he-gen wine.glass-acc break-past-conj Kiswu-also kkaythuli-ess-ta. break-past-decl ‘Thayswu broke his wine glass, and Kiswu broke, too.’

237 (4) Myengswu, Hyenswu, Cwunswu-ka wuntong hwu swui-ko iss-ess-ta. Myengswu Hyenswu Cwunswu-nom exercise after rest-prog past-decl Myengswu-ka Myengswu-uy umlyoswu-lul masi-ess-ta. Hyenswu-to Myengswu-nom Myengswu-gen beverage-acc drink-past-decl Hyenswu-also Myengswu-uy umlyoswu-lul masi-ess-ta. Myengswu-gen beverage-acc drink-past-decl ‘Myengswu, Hyenswu, and Cwunswu were taking a rest after exercise. Myengswu drank Myengswu’s beverage. Hyenswu also drank Myengswu’s beverage.’ Myengswu-ka ku-uy umlyoswu-lul masi-ess-ko, Hyenswu-to [e] Myengswu-nom he-gen beverage-acc drink-past-conj Hyenswu-also masi-ess-ta. drink-past-decl ‘Myengswu drank his beverage, and Hyenswu drank, too.’

(5) Pyengswu, Mwunswu, Cwunswu-ka cip-ul chengsoha-ko iss-ess-ta. Pyengswu Mwunswu Cwunswu-nom house-acc clean-prog past-decl Pyengswu-ka Pyengswu-uy pang-ul ssul-ess-ta. Mwunswu-to Pyengswu-nom Pyengswu-gen room-acc sweep-past-decl Mwunswu-also Pyengswu-uy pang-ul ssul-ess-ta. Pyengswu-gen room-acc sweep-past-decl ‘Pyengswu, Mwunswu, and Cwunswu were cleaning their house. Pyengswu swept Pyengswu’s room. Mwunswu also swept Pyengswu’s room.’ Pyengswu-ka ku-uy pang-ul ssul-ess-ko, Mwunswu-to [e] Pyengswu-nom he-gen room-acc sweep-past-conj Mwunswu-also ssul-ess-ta. sweep-past-decl ‘Pyengswu swept his room, and Mwunswu swept, too.’

(6) Chelswu, Yengswu, Cengswu-ka tosekwuan-eyse kongpwuha-ko iss-ess-ta. Chelswu Yengswu Cengswu-nom library-in study-prog past-decl Chelswu-ka Chelswu-uy pheyn-ul ttelethuli-ess-ta. Yengswu-to Chelswu-nom Chelswu-gen pen-acc drop-past-decl Yengswu-also Chelswu-uy pheyn-ul ttelethuli-ess-ta. Chelswu-gen pen-acc drop-past-decl ‘Chelswu, Yengswu, and Cengswu were studying in a library. Chelswu dropped Chelswu’s pen. Yengswu also dropped Chelswu’s pen.’ Chelswu-ka ku-uy pheyn-ul ttelethuli-ess-ko, Yengswu-to [e] Chelswu-nom he-gen pen-acc drop-past-conj Chelswu-also ttelethuli-ess-ta. drop-past-decl ‘Chelswu dropped his pen, and Yengswu dropped, too.’

238 (7) Chelho, Phengho, Swuho-ka kakca phyenci-lul ssu-ko iss-ess-ta. Chelho Phengho Swuho-nom each letter-acc write-prog past-decl Chelho-ka Chelho-uy phyenci-lul ilk-ess-ta. Phengho-to Chelho-nom Chelho-gen letter-acc read-past-decl Phengho-also Chelho-uy phyenci-lul ilk-ess-ta. Chelho-gen letter-acc read-past-decl ‘Chelho, Phengho, and Swuho were each writing a letter. Chelho read Chelho’s letter. Phengho also read Chelho’s letter.’ Chelho-ka ku-uy phyenci-lul ilk-ess-ko, Phengho-to [e] Chelho-nom he-gen letter-acc read-past-conj Phengho-also ilk-ess-ta. read-past-decl ‘Chelho read his letter, and Phengho read, too.’

(8) Kwangho, Changho, Yengho-ka ilsiktang-eyse cemsim-ul mek-ko Kwangho Changho Yengho-nom Japanese.restaurant-at lunch-acc eat-prog iss-ess-ta. Kwangho-ka Kwangho-uy utong-ul mek-ess-ta. past-decl Kwangho-nom Kwangho-gen Udon-acc eat-past-decl Changho-to Kwangho-uy utong-ul mek-ess-ta. Changho-also Kwangho-gen Udon-acc eat-past-decl ‘Kwangho, Changho, and Yengho were having lunch at a Japanese restaurant. Kwangho ate Kwangho’s Udon. Changho also ate Kwangho’s Udon.’ Kwangho-ka ku-uy utong-ul mek-ess-ko, Changho-to [e] Kwangho-nom he-gen Udon-acc eat-past-conj Changho-also mek-ess-ta. eat-past-decl ‘Kwangho ate his Udon, and Changho ate, too.’

(9) Pyengmin, Cengmin, Tongmin-ika wuntongcang-eyse chwukkwu-lul ha-ko Pyengmin Cengmin Tongmin-nom play.ground-in soccer-acc do-prog iss-ess-ta. Pyengmin-ika Pyengmin-iuy kong-ul cha-ss-ta. past-decl Pyengmin-nom Pyengmin-gen ball-acc kick-past-decl Cengmin-ito Pyengmin-iuy kong-ul cha-ss-ta. Cengmin-also Pyengmin-gen ball-acc kick-past-decl ‘Pyengmin, Cengmin, and Tongmin were playing soccer in a playground. Pyeng- min kicked Pyengmin’s ball. Cengmin also kicked Pyengmin’s ball.’ Pyengmin-ika ku-uy kong-ul cha-ss-ko, Cengmin-ito [e] Pyengmin-nom he-gen ball-acc kick-past-conj Cengmin-also cha-ss-ta. kick-past-decl ‘Pyengmin kicked his ball, and Cengmin kicked, too.’

239 (10) Cinswu, Sungswu, Pyengswu-ka seythaksil-eyse seythak-ul ha-ko Cinswu Sungswu Pyengswu-nom laundry.room-at laundry-acc do-prog iss-ess-ta. Cinswu-ka Cinswu-uy seycey-lul sayongha-yss-ta. past-decl Cinswu-nom Cinswu-gen detergent-acc use-past-decl Sungswu-to Cinswu-uy seycey-lul sayongha-yss-ta. Sungswu-also Cinswu-gen detergent-acc use-past-decl ‘Cinswu, Sungswu, and Pyengswu were doing laundry at a laundry room. Cinswu used Cinswu’s detergent. Sungswu also used Cinswu’s detergent.’ Cinswu-ka ku-uy seycey-lul sayongha-yss-ko, Sungswu-to [e] Cinswu-nom he-gen detergent-acc use-past-conj Sungswu-also sayongha-yss-ta. use-past-decl ‘Cinswu used his detergent, and Sungswu used, too.’

(11) Kimkenmo, Kimcanghwun, Sinsunghwun-i nolaypang-eyse nol-ko Kimkenmo Kimcanghwun Sinsunghwun-nom singing.room-at hang.out-prog iss-ess-ta. Kimkenmo-ka Kimkenmo-uy hithukok-ul pwulu-ess-ta. past-decl Kimkenmo-nom Kimkenmo-gen hit.song-acc sing-past-decl Kimcanghwun-to Kimkenmo-uy hithukok-ul pwulu-ess-ta. Kimcanghwun-also Kimkenmo-gen hit.song-acc sing-past-decl ‘Kimkenmo, Kimcanghwun, and Sinsunghwun were hanging out in a singing room. Kimkenmo sang Kimkenmo’s hit song. Kimcanghwun also sang Kimkenmo’s hit song.’ Kimkenmo-ka ku-uy hithukok-ul pwulu-ess-ko, Kimcanghwun-to [e] Kimkenmo-nom he-gen hit.song-acc sing-past-conj Kimcanghwun-also pwulu-ess-ta. sing-past-decl ‘Kimkenmo sang his hit song, and Kimcanghwun sang, too.’

(12) Pongswu, Thaykang, Thayho-ka pwuek-eyse selkeci-lul ha-ko Pongswu Thaykang Thayho-nom kitchen-at dish.washing-acc do-prog iss-ess-ta. Pongswu-ka Pongswu-uy khep-ul sis-ess-ta. past-decl Pongswu-nom Pongswu-gen cup-acc wash-past-decl Thaykang-ito Pongswu-uy khep-ul sis-ess-ta. Thaykang-also Pongswu-gen cup-acc wash-past-decl ‘Pongswu, Thaykang, and Thayho were doing the dishes at a kitchen. Pongswu washed Pongswu’s cup. Thaykang also washed Pongswu’s cup.’ Pongswu-ka ku-uy khep-ul sis-ess-ko, Thaykang-ito [e] Pongswu-nom he-gen cup-acc wash-past-conj Thaykang-also sis-ess-ta. wash-past-decl ‘Pongswu washed his cup, and Thaykang washed, too.’

240 (13) Chiswu, Pyengswu, Hanswu-ka kkaphey-eyse tayhwua-lul nanwu-ko Chiswu Pyengswu Hanswu-nom café-at conversation-acc share-prog iss-ess-ta. Chiswu-ka Chiswu-uy cen yecachinkwu-lul past-decl Chiswu-nom Chiswu-gen previous girlfriend-acc helttut-ess-ta. Pyengswu-to Chiswu-uy cen yecachinkwu-lul speak.ill.of-past-decl Pyengswu-also Chiswu-gen previous girlfriend-acc helttut-ess-ta. speak.ill.of-past-decl ‘Chiswu, Pyengswu, and Hanswu were having a conversation at a café. Chiswu spoke ill of Chiswu’s ex-girlfriend. Pyengswu also spoke ill of Chiswu’s ex- girlfriend.’ Chiswu-ka ku-uy cen yecachinkwu-lul helttut-ess-ko, Chiswu-nom he-gen previous girlfriend-acc speak.ill.of-past-conj Pyengswu-to [e] helttut-ess-ta. Pyengswu-also speak.ill.of-past-decl ‘Chiswu spoke ill of his ex-girlfriend, and Pyengswu spoke ill of, too.’

(14) Congswu, Hwuanswu, Tongswu-ka kongwuen-eyse nol-ko Congswu Hwuanswu Tongswu-nom park-at hang.out-prog iss-ess-ta. Congswu-ka Congswu-uy kay-lul kancilephi-ess-ta. past-decl Congswu-nom Congswu-gen dog-acc tickle-past-decl Hwuanswu-to Congswu-uy kay-lul kancilephi-ess-ta. Hwuanswu-also Congswu-gen dog-acc tickle-past-decl ‘Congswu, Hwuanswu, and Tongswu were hanging out at a park. Congswu tickled Congswu’s dog. Hwuanswu also tickled Congswu’s dog.’ Congswu-ka ku-uy kay-lul kancilephi-ess-ko, Hwuanswu-to [e] Congswu-nom he-gen dog-acc tickle-past-conj Hwuanswu-also kancilephi-ess-ta. tickle-past-decl ‘Congswu tickled his dog, and Hwuanswu tickled, too.’

(15) Tongwu, Cinwu, Thaywu-ka Sinswu-uy cip-eyse nol-ko Tongwu Cinwu Thaywu-nom Sinswu-gen house-at hang.out-prog iss-ess-ta. Tongwu-ka Tongwu-uy yetongsayng-ul ttayli-ess-ta. past-decl Tongwu-nom Tongwu-gen younger.sister-acc hit-past-decl Cinwu-to Tongwu-uy yetongsayng-ul ttayli-ess-ta. Cinwu-also Tongwu-gen younger.sister-acc hit-past-decl ‘Tongwu, Cinwu, and Thaywu were hanging out at Sinswu’s house. Tongwu hit Tongwu’s younger sister. Cinwu also hit Tongwu’s younger sister.’ Tongwu-ka ku-uy yetongsayng-ul ttayli-ess-ko, Cinwu-to [e] Tongwu-nom he-gen younger.sister-acc hit-past-conj Cinwu-also ttayli-ess-ta. hit-past-decl

241 ‘Tongwu hit his younger sister, and Cinwu hit, too.’

(16) Wonmwu, Kiswu, Sengswu-ka chatcip-eyse tayhwua-lul nanwu-ko Wonmwu Kiswu Sengswu-nom tea.house-at conversation-acc share-prog iss-ess-ta. Wonmwu-ka Wonmwu-uy nwuna-lul past-decl Wonmwu-nom Wonmwu-gen older.sister-acc chingchanha-yss-ta. Kiswu-to Wonmwu-uy nwuna-lul compliment-past-decl Kiswu-also Wonmwu-gen older.sister-acc chingchanha-yss-ta. compliment-past-decl ‘Wonmwu, Kiswu, and Sengswu were having a conversation at a tea house. Won- mwu complimented Wonmwu’s older sister. Kiswu also complimented Wonmwu’s older sister.’ Wonmwu-ka ku-uy nwuna-lul chingchanha-yss-ko, Kiswu-to [e] Wonmwu-nom he-gen older.sister-acc compliment-past-conj Kiswu-also chingchanha-yss-ta. compliment-past-decl ‘Wonmwu complimented his older sister, and Kiswu complimented, too.’

F.3 Condition 3: Quantiﬁcational-Bound

The test items for the Quantificational-Bound condition are the same as those for the Quantificational-Bound condition condition in Experiment 3.

F.4 Condition 4: Quantiﬁcational-Free

The test items for the Quantificational-Free condition are the same as those for the Quantificational-Free condition condition in Experiment 3.

242